World Health Organization Recommendations on Caregiving

World Health Organization Recommendations on Caregiving Interventions to

Support Early Child Development in the First Three Years of Life:

Report of the systematic review of evidence

Joshua Jeong PhD, Emily Franchett MSc, Aisha K. Yousafzai PhD

Department of Global Health and Population

Harvard T.H. Chan School of Public Health

12 November 2018

(finalized 08 Feb 2020)

Contact information:

Aisha K. Yousafzai, PhD

Associate Professor

Department of Global Health and Population

Harvard T.H. Chan School of Public Health

665 Huntington Avenue, Boston, MA 02115, USA

Email: [email protected]

mailto:[email protected]

2 | P a g e

Executive summary

An estimated 250 million children are at risk of not achieving their developmental

potential in the first 5 years of life. Policies and programmes that enable caregivers to support

young children’s development are critical. Evidence in the series Early childhood development:

from science to scale published in the Lancet in 2017 highlighted the importance of nurturing

care comprising caregivers’ practices with respect to health, nutrition, safety and security,

responsive caregiving, and early learning opportunities that support children’s healthy

development. The health sector plays a critical role in delivering key nurturing care

interventions, and there is opportunity to strengthen the nurturing care approach in health

services and in partnership with other sectors. In order to guide the health sector, the World

Health Organization (WHO) intends to develop evidence-based guidelines for approaches to

improve early childhood development (ECD) enabling Member States to make informed

decisions about a range of policy and programme actions in their efforts to achieve targets in the

Sustainable Development Goals pertaining to health, learning and behaviour for human

development. These guidelines will complement and support existing WHO guidelines and tools

relevant to nurturing care.

A Guideline Development Group (GDG) comprising experts in the field of ECD and

health proposed four1 key research questions for review of evidence to inform recommendations

relating to early learning and responsive care.

1. What is the effectiveness of responsive caregiving on ECD in the first 3 years of life?

2. What is the effectiveness of caregiving interventions that promote early learning on ECD

in the first 3 years of life?

3. What is the effectiveness of caregiving to support healthy child socioemotional and

behavioural development on ECD in the first 3 years of life?

4. What are the effects of combined caregiving and nutrition programmes on ECD and child

growth outcomes in the first 3 years of life?

In many interventions, strategies to promote early learning and responsive caregiving were

combined; therefore, it was agreed to carry out additional analysis related to the first two

questions:

2.1 What is the effectiveness of caregiving interventions that combine both responsive

caregiving and the promotion of early learning on ECD in the first 3 years of life?

1 A fifth question was addressed in a separate review.

3 | P a g e

2.2. What is the effectiveness of any caregiving interventions (responsive caregiving, promotion

of early learning, or combined responsive caregiving and the promotion of early learning) on

ECD in the first 3 years of life.

This report is a synthesis of the evidence undertaken by a systematic review team at the

Harvard T.H. Chan School of Public Health. In total, across the four original questions we report

on 67 studies (77 records): question 1 (n = 17 studies, 19 records), question 2 (n = 22 studies, 22

records), question 3 (n = 10 studies, 11 records), and question 4 (n = 18 studies, 25 records),

were identified for review. Outcomes assessed included ECD, attachment, child behaviour, child

growth, child health and nutrition, caregiving knowledge, caregiving practices, caregiver-child

interactions, and caregivers’ mental health. Quality of evidence was assessed using the Grading

of Recommendations Assessment, Development and Evaluation (GRADE) approach. The final

recommendations based on the quality of the evidence will be determined by the GDG informed

by evidence-to-decision making tables for the key research questions.

4 | P a g e

Acknowledgments

We are grateful to the members of the systematic review team at the Harvard T.H. Chan

School of Public Health for their contributions to reviewing studies and synthesizing evidence

for analysis: Alastair Fung, Raghbir Kaur, Helen Pitchik, Vijayaragavan Prabakaran, Clariana

Vitória Ramos, Mathilda Regan, Katharine Robb, Rucha Shelgikar, and Yu-Cheng Tsai. We

thank Carol Mita, librarian at the Francis A. Countway Library of Medicine, Harvard Medical

School and Boston Medical Library, for her assistance in the literature search. We also thank the

GDG chaired by Jane Fisher, Monash University, for its contributions to the process of the

review of evidence. The support and guidance of the WHO team, including Bernadette

Daelmans, Tarun Dua and Nigel Rollins, is gratefully acknowledged.

5 | P a g e

Table of contents

Executive summary ......................................................................................................................... 2

Acknowledgments........................................................................................................................... 4

Introduction and scope of the review .............................................................................................. 7

Methods......................................................................................................................................... 12

Evidence and recommendations for responsive caregiving interventions (PICO 1) .................... 22

Evidence and recommendations for caregiving interventions that promote early learning (PICO

2) ................................................................................................................................................... 26

Evidence and recommendations for caregiving interventions that combine responsive caregiving

and promotion of early learning: additional analysis 2.1 .............................................................. 31

Evidence and recommendations for all caregiving interventions that deliver responsive

caregiving, promotion of early learning, or combined responsive caregiving and promotion of

early learning: Additional analysis 2.2 ......................................................................................... 36

Evidence and recommendations for caregiving interventions to support healthy child

socioemotional and behavioural development (PICO 3) .............................................................. 39

Evidence and recommendations for combined caregiving and nutrition interventions (PICO 4) 44

Research gaps................................................................................................................................ 58

References ..................................................................................................................................... 59

Appendix A: Glossary................................................................................................................... 69

Appendix B: List of Guideline Development Group Members .................................................... 74

Appendix C: Summary of interventions for all included studies in the systematic reviews ........ 75

Appendix D: GRADE tables and analysis for responsive caregiving interventions (n=17) ........ 76

Appendix E: GRADE tables and analysis for caregiving interventions to support early learning

opportunities (n=22) ..................................................................................................................... 85

Appendix F: GRADE tables and analysis for combined responsive caregiving and the promotion

of early learning interventions (n=42) .......................................................................................... 94

Appendix G: GRADE tables and analysis for caregiving interventions for responsive caregiving,

early learning, or a combined responsive caregiving and promotion of early learning intervention

(n=81) .......................................................................................................................................... 103

Appendix H: GRADE tables and analysis for caregiving interventions to support healthy child

socioemotional and behavioural development (n=10) ................................................................ 112

Appendix I: GRADE tables and analysis for integrated caregiving and nutrition interventions

(n=18) .......................................................................................................................................... 117

6 | P a g e

Abbreviations and acronyms2

BSID Bayley Scale of Infant and Toddler Development

CBCL Child Behavior Check List

CES-D Center for Epidemiological Studies Depression Scale

CI Confidence interval

c-RCT Cluster randomized controlled trial

ECD Early childhood development

ES Effect size

GDG Guideline Development Group

GRADE Grading of Recommendations Assessment, Development and Evaluation

HAZ Height-for-age Z-score

HIC High-income country

HOME Home Observation for Measurement of the Environment

ITSEA Infant-Toddler Social Emotional Assessment

LMIC Low- and middle-income country

NCAST Nursing Child Assessment Satellite Teaching

PICO Population/Problem, Intervention, Comparison/Control, Outcome

PROSPERO Prospectively registered systematic reviews in health and social care, welfare,

public health, education, crime, justice, and international development

RCT Randomized controlled trial

SDQ Strengths and Difficulties Questionnaire

WAZ Weight-for-age Z-score

WHO World Health Organization

WHZ Weight-for-height Z-score

2 See Appendix A for Glossary of terms.

7 | P a g e

Introduction and scope of the review

An estimated 250 million children are at risk of not achieving their developmental

potential in the first 5 years of life (Lu et al., 2016). Policies and programmes that enable

caregivers to support young children’s development are critical. Evidence in the series Early

childhood development: from science to scale published in the Lancet highlighted the importance

of nurturing care comprising caregiver and family practices in health, nutrition, safety and

security, responsive caregiving, and early learning opportunities that support children’s healthy

development (Figure 1) (Black et al., 2017; Britto et al., 2017; Richter et al., 2017). A central

tenant of the nurturing care concept is the process of care encompassing caregiving practices

such as caregiver sensitivity to children’s physical and emotional needs, protection from harm,

provision of opportunities for exploration and learning, and interactions with young children that

are responsive, emotionally supportive, and cognitively stimulating.

Figure 1. Domains of nurturing care (Black et al., 2017).

While all sectors have roles and responsibilities in promoting policies and programmes

for nurturing care, the health sector plays a particularly important role during the first 3 years of

life (from conception to 2 years of age). Firstly, the health sector can reach the youngest

children and their caregivers in the first 3 years of life to promote protective care and mitigate

risk factors in an especially important and sensitive period of brain development (Nelson, 2000).

Secondly, health services commonly implement components of nurturing care (e.g. breastfeeding

promotion or the Integrated Management of Childhood Illness) that reduce risks for childhood

mortality and morbidity as well as support early development but could potentially be

strengthened with a holistic approach to the provision of nurturing care (Aboud & Yousafzai,

8 | P a g e

2016; Britto et al., 2017). Thirdly, the health sector can play a strong role in leveraging

partnerships across sectors that enable the coordination of services for young children (Richter et

al., 2017).

In order to guide the health sector, WHO intends to develop evidence-based guidelines

for approaches to improve ECD enabling Member States to make informed decisions about a

range of policy and programme actions in their efforts to achieve targets in the Sustainable

Development Goals pertaining to health, learning and behaviour for human development. These

guidelines will complement and support existing WHO guidelines and tools relevant to nurturing

care (Table 1).

Table 1. WHO guidelines and tools that support nurturing care

Nurturing care domain Guidelines and tools

Nutrition, Health

Focus on infant and young

child feeding, care for the

newborn

Courses

• Infant and young child feeding counselling: an integrated course

• Combined course on growth assessment and infant and young child feeding

counselling

• Integrated Management of Childhood Illness

• Essential newborn care

• Caring for newborns and children in the community: a training course for

community health workers (including modules on caring for the newborn at

home, caring for the sick child and caring for the child‘s healthy growth

and development)

Guidelines

• Optimal feeding of low-birth-weight infants in low- and middle-income

countries

• Guidelines on the management of children with severe acute malnutrition

• Protecting, promoting and supporting breastfeeding in facilities providing

maternity and newborn services (includes updates on the Ten steps to

successful breastfeeding)

• HIV and infant feeding: framework for priority action

Health

Focus on pregnancy, safe

delivery and perinatal care

Guidelines

• Recommendations on antenatal care for a positive pregnancy experience

• Companion of choice during labour and childbirth for improved quality of

care

• Prevention and treatment of maternal peripartum infections

• Interventions to improve preterm birth outcomes

• Postnatal care of the mother and newborn

Health, Security and Safety

Focus on protection of young

children, and well-being of

children with disabilities

Guidelines

• INSPIRE: seven strategies for ending violence against children

• WHO Global disability action plan 2014-2021

• Ten strategies for keeping children safe on the road

• Early childhood development and disability

Health, Security and Safety

Focus on water, sanitation and

hygiene interventions

Guidelines

• The impact of the environment on children’s health

• Investing in water and sanitation: increasing access, reducing inequalities

• Progress on drinking water, sanitation and hygiene

9 | P a g e

While these guidelines and tools support ECD and highlight the importance of the nurturing care

domains of responsive caregiving and early learning, there is a need to develop focused

guidelines to better integrate a range of caregiving interventions in health and other services.

Aims and objectives of the guideline

The aim of the guideline is to improve ECD. The three objectives of are to:

1. identify ECD-specific interventions that are effective in improving developmental

outcomes in children;

2. identify effective, feasible approaches to deliver interventions to improve ECD outcomes;

3. consolidate in one guideline WHO-recommended interventions that promote ECD.

The target audience for the guideline includes district and sub-national health managers, health

workers, development agencies and implementing partners, nongovernmental organizations and

policy-makers working in the area of maternal and child health.

The guideline will be developed following the process outlined in the WH O Handbook

for guideline development, second edition. The Departments of Maternal, Newborn, Child and

Adolescent Health and Mental Health and Substance Abuse have identified experts for the GDG.

A scoping meeting was convened in September 2017 to formulate questions to be addressed in

systematic reviews to inform the recommendations of the GDG (see Appendix B for list of GDG

members).

Scope of the systematic review

The purpose of the systematic reviews is to determine the level of evidence available to

support potential recommendations for Member States on the promotion of ECD. This

systematic review presents evidence for a total of six questions (four from the original scoping

process, and two additional analyses requested later) formulated using the Population/Problem,

Intervention, Comparison/Control, and Outcome (PICO) framework (Table 2).

Table 2. PICO questions for systematic review

PICO # Research question Population Intervention Comparison Outcomes

1 What is the effectiveness of

responsive caregiving

interventions on ECD in the

first 3 years of life?

Conception

to 3 years of

life

Global

Caregiving

interventions that

only implement

responsive

caregiving

Standard of

care or

control

Primary: ECD

Other: child

growth, child

nutrition, child

health,

caregiving,

caregiver mental

health


caregiving interventions that

promote early learning on

Conception

to 3 years of

life

Caregiving

interventions that

only support early

Standard of

care or

control

Primary: ECD

Other: child

growth, child

10 | P a g e

PICO # Research question Population Intervention Comparison Outcomes

ECD in the first 3 years of

life?

Global

learning and

development

nutrition, child

health,

caregiving,

caregiver mental

health

2.1

(additional

analysis)

What is the effectiveness of

caregiving interventions that

combine both responsive

caregiving and the promotion

of early learning on ECD in

the first 3 years of life?

Conception

to 3 years of

life

Global

Caregiving

interventions with

components of

both responsive

caregiving and the

promotion of

early learning

Standard of

care or

control

Primary: ECD

Other: child

attachment, child

growth, child

nutrition, child

health,

caregiving,

caregiver mental

health

2.2

(additional

analysis)

What is the effectiveness of

any caregiving interventions

(responsive caregiving,

promotion of early learning,

or combined responsive

caregiving and the promotion

of early learning) on ECD in

the first 3 years of life?

Conception

to 3 years of

life

Global

All interventions

from PICOs 1, 2,

and 2.1 are

included

Standard of

care or

control

Primary: ECD

Other: child

attachment, child

growth, child

nutrition, child

health,

caregiving,

caregiver mental

health


caregiving to support healthy

child socioemotional and

behavioural development on

ECD in the first 3 years of

life?

Conception

to 3 years of

life

Global

Caregiving to

support healthy

socioemotional

and behavioural

development

Standard of

care or

control

Primary: ECD

Other: child

growth, child

nutrition, child

health,

caregiving,

caregiver mental

health

4 What are the effects of

combined caregiving and

nutrition programmes on

ECD and child growth

outcomes in the first 3 years

of life?

• What are the

independent and additive

effects of caregiving and

nutrition interventions on

ECD and child growth

outcomes in the first 3

years of life?

• Do the effects on ECD

and child growth

outcomes differ between

programmes that are

targeted for young

children with moderate

to severe malnutrition

compared to general

programmes?

Conception

to 3 years of

life

Global

Combined

caregiving and

nutrition

interventions

Standard of

care or

control

Primary: ECD,

child growth

Other: child

attachment, child

nutrition, child

health,

caregiving,

caregiver mental

health

11 | P a g e

The components of nurturing care interventions addressed in this review are responsive

caregiving and support for early learning and development. We organized the analysis by

whether the intervention focuses only on one component or whether it combines both

components (i.e. responsive caregiving and support for early learning and development), which

we reference as PICO questions 1, 2 and additional analysis 2.1 and 2.2 respectively.

A challenge in the literature is the wide application of definitions of the nurturing care

components (responsive caregiving and support for early learning and development), which are

not consistently operationalized in the studies. Therefore, a central aspect to the review process

was to establish a clear definition for each nurturing care component, which iteratively evolved

over the first weeks of the data extraction process as the team reviewed and discussed each

intervention. This categorization is broad and is based on often limited descriptions reported

about the interventions in the published studies; however, a degree of delineation was feasible.

The purpose of the additional analysis (2.2) which includes all the caregiving interventions

together by merging intervention studies under PICO 1, 2 and 2.1 was to acknowledge that broad

categorization and the potential for overlap across intervention strategies. A general conclusion

may be drawn about the effect of general caregiving interventions to support ECD from the

additional analysis.

However, the PICO questions were not framed to ask whether a specific caregiving

component had greater or lesser effects on ECD than another, and studies have not been designed

to address this question. Thus, the purpose of the additional analysis was not to compare the

different components (i.e. support for early learning and responsive caregiving). It is

recommended that any comparisons of caregiving components are interpreted with caution given

the wide range of applications of these concepts, the diverse approaches to combining these

concepts and the variation in the implementation characteristics for the interventions.

12 | P a g e

Methods

This systematic review protocol was developed in accordance with the process outlined

in the WHO Handbook for guidelines development, second edition. A unique protocol was

developed for the PICO questions and registered with Prospectively registered systematic

reviews in health and social care, welfare, public health, education, crime, justice, and

international development (PROSPERO) (https://www.crd.york.ac.uk/PROSPERO) (Table 3) .

Table 3. PICO registration on PROSPERO

PICO # Research question Protocol registration ID

1 What are the effects of responsive caregiving interventions on ECD in the


CRD42018092458 and

CRD42018092461

2 What are the effects of caregiving interventions that promote early learning

on ECD in the first 3 years of life?

2.1 What is the effectiveness of caregiving interventions that combine both

responsive caregiving and the promotion of early learning on ECD in the


2.2 What is the effectiveness of any caregiving interventions (responsive

caregiving, promotion of early learning, or combined responsive caregiving

and the promotion of early learning) on ECD in the first 3 years of life?

3 What are the effects of caregiving to support healthy child socioemotional

and behavioural development on ECD in the first 3 years of life?

CRD42018092462

4 What are the effects of integrated caregiving and nutrition programmes on

ECD and child growth outcomes in the first 3 years of life?

• What are the independent and additive effects of caregiving and

nutrition interventions on ECD and child growth outcomes in the first

3 years of life?

• Do the effects on ECD and child growth outcomes differ between

programmes that are targeted for young children with moderate to

severe malnutrition compared to general programmes?

CRD42018092605

The reviews, analysis and report were completed by a team at the Harvard T.H. Chan School of

Public Health between 15 January and 5 September 2018 (including a preliminary analysis and

report reviewed by the GDG in May 2018).

Inclusion and exclusion criteria for the systematic reviews

The systematic reviews were subject to a common set of inclusion/exclusion criteria

across the PICO questions, and a specific set of inclusion and exclusion criteria for each of the

different questions. No temporal or regional constraints were imposed.

Inclusion criteria:

• Primary studies from peer-reviewed journals.

https://www.crd.york.ac.uk/PROSPERO

13 | P a g e

• Interventions targeted toward caregivers, which we define using a modified operational

definition of a caregiving programme from UNICEF as one that incorporates “activities,

programmes, services or interventions, for caregivers, aimed at improving caregiver

interaction, behaviors, knowledge, beliefs, attitudes and practices” (Britto et al., 2015).

Caregivers were defined as the legal guardian, biological parent, or adult responsible for the

well-being of the child.

• Interventions that were evaluated using a randomized controlled study design with at least

one control group and one intervention group.

• Interventions that targeted children and their caregivers in the early life course (pregnancy

through the first 3 years of life).

• Interventions that assessed at least one measure of ECD (cognition, language, motor,

socioemotional development) or behaviour as a primary outcome.

• Evaluations that assessed outcomes immediately after the completion of the intervention (or

shortly thereafter).

Exclusion criteria:

• Interventions that targeted children who were preterm or who have a chronic illness, very

low birth weight, or disability.3

• Interventions conducted with caregivers who have an illness or disability.

• Interventions that were not evaluated using a randomized controlled study design.

• Interventions that were not relevant to caregiving.

• Interventions that did not assess at least one ECD outcome.

• Interventions that enrolled children older than 3 years of age (on average).

Literature search and screening

The research team worked closely with a research librarian with expertise in

comprehensive bibliographic database searching. Intervention studies were searched across five

electronic bibliographic databases: EMBASE, PubMed, PsycINFO, Cochrane Central Register of

Controlled Trials (CENTRAL), and ERIC. Search strategies were developed in accordance with

each database. A string of keywords was determined to capture four broad categories: (1)

caregiving interventions, (2) targeting children aged from pregnancy through 3 years, (3) that

were evaluated using a randomized controlled study design, and (4) assessed an ECD outcome.

Keywords were informed by search terms and keywords used in prior systematic reviews related

to caregiving interventions (Aboud & Yousafzai, 2015; Britto et al., 2017; Eshel et al., 2006;

Mol et al., 2008), as well as through consultations with the research librarian to select relevant

MESH terms and search filters. The full list of key search terms is shown in Box 1.

3 Interventions targeting children with low birth weight (< 2500g) were included and interventions targeting children

with very low birth weight alone excluded (< 1500g).

14 | P a g e

Box 1: Key search terms

(Infant[Mesh] OR "Child, Preschool"[Mesh] OR infant[tiab] OR infants[tiab] OR infant’s[tiab] OR neonate[tiab]

OR neonates[tiab] OR neonatal[tiab] OR newborn*[tiab] OR new born*[tiab] OR baby[tiab] OR babies[tiab] OR

toddler[tiab] OR toddlers[tiab] OR toddlerhood[tiab] OR preschool*[tiab] OR pre school*[tiab] OR early

childhood[tiab] OR young children*[tiab] OR "Perinatal Care"[Mesh] OR perinatal[tiab] OR antenatal[tiab] OR

ante natal[tiab] OR postnatal[tiab] OR post natal[tiab] OR age 0[tiab] OR aged 0[tiab] OR age zero[tiab] OR aged

zero[tiab] OR age 1[tiab] OR aged 1[tiab] OR age one[tiab] OR aged one[tiab] OR age 2[tiab] OR aged 2[tiab] OR

age two[tiab] OR aged two[tiab] OR 1 year old*[tiab] OR one year old*[tiab] OR 2 year old*[tiab] OR two year

old*[tiab] OR 3 year old*[tiab] OR three year old*[tiab] OR 3 years of age[tiab] OR 2 years of age[tiab] OR 1 year

of age[tiab] OR under 2 years[tiab] OR under 1 year[tiab]) AND ("Parenting"[Mesh] OR "Child Rearing"[Mesh]

OR "Maternal Behavior"[Mesh] OR "Parent-Child Relations"[Mesh] OR "Parents"[Mesh] OR "Caregivers"[Mesh]

OR parents[tiab] OR parenting[tiab] OR mother[tiab] OR mothers[tiab] OR maternal behav*[tiab] OR parental

behav*[tiab] OR paternal behavior[tiab] OR parent infant[tiab] OR infant parent[tiab] OR father[tiab] OR

fathers[tiab] OR caregiv*[tiab] OR care giv*[tiab] OR child rearing[tiab]) OR parent child[tiab] OR child

parent[tiab] OR parent training[tiab] OR parent education[tiab] OR parental training[tiab] OR parental

education[tiab] AND ("Child Behavior"[Mesh] OR "Child Development"[Mesh] OR "Cognition"[Mesh] OR

"Executive Function"[Mesh] OR "Emotional Intelligence"[Mesh] OR "Emotions"[Mesh] OR "Motor Skills"[Mesh]

OR attachment[tiab] OR attention[tiab] OR behavior[tiab] OR behavioral[tiab] OR behaviors[tiab] OR

behaviour[tiab] OR behavioural[tiab] OR behaviours[tiab] OR child development[tiab] OR cognition[tiab] OR

cognitive[tiab] OR communication[tiab] OR communicative[tiab] OR compliance[tiab] OR conduct problem*[tiab]

OR executive function*[tiab] OR emotional[tiab] OR emotions[tiab] OR empathy[tiab] OR fine motor[tiab] OR

language[tiab] OR mastery[tiab] OR motivation[tiab] OR motor skill*[tiab] OR peer relation*[tiab] OR play

skills[tiab] OR prosocial[tiab] OR reading[tiab] OR social[tiab] OR socialization[tiab] OR socio emotion*[tiab] OR

socioemotion*[tiab] AND random*[tw] AND English[lang]

Reference lists of relevant studies and reviews were also perused for any additional studies that

were not retrieved from the search strategy.

The screening process was completed by authors EF and JJ with support from AKY and

another member of the research team experienced in systematic reviews (HP). Initial screening

of titles and abstracts was completed in accordance with the inclusion and exclusion criteria.

Next, full texts of selected studies were reviewed to determine eligibility and specific reasons

were documented for excluding articles at this stage. In all cases, two systematic review team

members independently assessed the eligibility of each study to ensure accuracy. Discrepancies

were resolved through group discussion with input from author AKY.

Studies that were selected for data extraction were assigned to a graduate research assistant

who extracted data using a standardized, pre-piloted form. Quantitative and qualitative data that

were extracted included:

• details related to sample characteristics;

• details related to intervention characteristics;

• details related to risk of bias;

• data necessary for effect size calculation for all outcomes of interest (means, standard

deviations, sample sizes of pre- and post-scores in intervention and control groups).

15 | P a g e

All data extraction team members were trained to use the data extraction sheet and received a

detailed standard operating protocol that included a set of standardized definitions for

intervention characteristics. EF held weekly in-person meetings with all data extractors to

monitor team progress and address any discrepancies that arose in the data extraction process.

All data extraction forms were reviewed by EF and JJ on a weekly basis. For additional quality

assurance, 50% of included articles were subsampled for quality assurance by either EF or JJ

during the first three weeks of data extraction while research assistants were familiarizing

themselves with the process. During the remaining weeks of the data extraction process, 20% of

studies were subsampled for quality assurance by EF, JJ or AKY.4 The flow of articles from the

original search through the final selected studies is shown using the Preferred Reporting Items

for Systematic Reviews and Meta-Analyses flow diagram (Figure 2).

The review was submitted mid-April 2018. Following a review of the preliminary analysis by the

GDG in May 2018, all excluded articles were rescreened by EF, JJ and AKY, and any additional

data extraction was managed by EF and JJ. This was to include interventions targeting children

with low birth weight and to include studies with a sample size smaller than 85 that had been

excluded in the original analysis. Responses to GDG reviews were submitted on Sept 5th 2018

following the May 2018 GDG meeting.

4Data extraction between June and August 2018.

16 | P a g e

Figure 2. Data screening and selection process5

5 A second search without English language restrictions was completed. This generated 439 unique studies which we

did not review due to resource limitations.

9,579 unique records identified

from search of 5 databases

11 additional records identified

from other sources

9,590 titles of records screened

923 records excluded

259 full-texts of records assessed for

eligibility 159 records excluded•62 - no ECD outcome

assessed

•41 – children are older than

age-3 years

•21 – not a caregiving

intervention

•18 – longer term follow-

up/secondary analysis of a

program

•17 - clinically targeted

population

•10 – not an RCT90 records included in systematic review

and meta-analysis

1,182 abstracts of records screened

8,408 records excluded

19 records in PICO 1

(n=17 studies)


(n=22 studies)


(n=42 studies)


(n=18 studies)

17 | P a g e

Assessment of risk of bias of individual studies

Each individual study selected for the review was assessed for risk of bias using the

criteria outlined by the Cochrane risk of bias assessment tool for randomized controlled trials

recommended by the Cochrane Handbook version 5.1.0. The risk of bias in the included studies

was assessed alongside the data extraction process by considering the following characteristics:

(1) randomization sequence generation; (2) concealment of allocation to treatment group; (3)

blinding of participants and investigators; (4) reporting of data on all study participants (taking

into account attrition and exclusions); (5) complete reporting of all study outcomes that were

specified a priori; (6) other sources of bias (including considerations around measurement

reliability and validity). Disagreements between the team members over the risk of bias

inparticular studies were resolved by group discussion with involvement of AKY where

necessary.

Synthesis, analysis, and reporting

A challenge in the literature is the wide application of definitions for caregiving (or

parenting) interventions, which are not consistently defined in the literature. Therefore, a central

aspect to the review process was to establish a clear definition for each type of intervention,

which iteratively evolved over the first weeks of the data extraction process as the team reviewed

and discussed each intervention (see Table 4 describing how each intervention was defined for

analysis).

A meta-analysis was conducted for each of the following outcomes: ECD, child

behaviour, child attachment, child growth, child nutrition, child health, caregiving (practices,

knowledge, and caregiver-child interactions), and caregiver mental health if the number of

available studies for the specific outcome was sufficient.

• Standardized tools were selected over non-standardized tools or study-specific tools.

• Comprehensive assessments were selected over screens or single items.

A predetermined decision-making matrix was developed for situations where a study reported

more than one result for a specific outcome of interest. This was considered the most

parsimonious approach as recommended by statistical advisors. The decision-making matrix was

informed by the developmental theory for language and motor development. Across the pool of

80 studies from which data were extracted, this applied to five studies reporting language

development and two studies reporting motor development. Given the small number of studies

dispersed across the PICO questions and the underlying theoretical rationale, a sensitivity

analysis was not conducted.

18 | P a g e

• For child language, receptive language was selected over expressive language (if a

combined score was not possible to ascertain due to the assessment tool used, the degree

of adaptation/modification to any tool or limitations in the data reported). Children

typically develop receptive language skills first, and this is especially important in the age

group of interest (i.e. 0-3 years) in this meta-analysis.

• For child motor development, fine motor was selected over gross motor (if a combined

score was not possible to ascertain due to the assessment tool used, the degree of

adaptation/modification to any tool or limitations in the data reported). There is evidence

in the literature indicating an association between the development of fine motor skills in

early life with subsequent learning and development.

For child behavioural problems, separate analyses were undertaken for internalizing and

externalizing problem behaviours. If multiple behavioural scores were reported in either category

which could not be separated using developmental theory, then the protocol adopted by Tanner-

Smith and Tipton (2014) was followed. The robust variance estimation allows adjustment for

correlated effect sizes between the different outcomes from the same study for a given domain.

In each PICO question, a sub-analysis was also conducted by regional context: globally,

high-income countries (HICs), and low- and middle-income countries (LMICs). PICOs 1 and 2

were predominantly studies from HICs, 2.1 and 2.2 were a mixture of both HICs and LMICs,

and PICOs 3 and 4 comprised studies only from HICs.

Quantitative data synthesis was conducted with either fixed or random effects meta-

analysis based on a test of heterogeneity. Heterogeneity was evaluated using Cochran's Q test of

heterogeneity and the I² statistic. Subgroup analyses were conducted to identify potential sources

of heterogeneity using random-effect meta-analysis methods to estimate the average effect across

groupings of studies. Results from the meta-analyses are presented with a forest plot. The Egger

regression asymmetry test for publication bias will be calculated for the outcomes to assess the

possibility of publication bias.

Additionally, EF, JJ and AKY summarized the quality of the body of evidence for each

outcome of interest listed in the questions, in accordance with the GRADE approach. GRADE

evidence profiles, containing the assessment of the quality of the evidence and a summary of

findings across studies for each important or critical outcome were created. The quality of the

body of evidence for each outcome was categorized as high, moderate, low or very low (see

Appendix C for a summary of each intervention study and Appendices D-I for GRADE profiles

for each question). In summary, across the questions we report on 81 studies: question 1 (n=17),

question 2 (n=22), additional analysis 2.1 (n=42), additional analysis 2.2 (81) question 3 (n=10)

and question 4 (n=18), were identified for review.

19 | P a g e

Table 4. Intervention definitions for PICO questions

Intervention Focus Description and Type of Interventions Types of Interventions

Excluded

Interventions that

only implement

responsive

caregiving

Responsive caregiving interventions target the

caregiver-child dyad to promote responsive

caregiver-child interactions and strengthen the

parent-child relationship. These interventions

encourage and support sensitivity and responsiveness

(care that is prompt, consistent, contingent, and

appropriate to the child’s cues, signals, behaviours

and needs) or secure attachment. Interventions that

improve caregivers’ abilities to incorporate the

child’s signals and perspective can be undertaken in

the context of, but not limited to, play and

communication or feeding. They include, but are not

limited to, facilitating the caregiver to be attuned to

and identify the child’s needs and wants, to follow

the child’s lead, help the child to focus, support the

child’s exploration and scaffold development

• Interventions that relate to

caregiving more generally,

but without a primary focus

on promoting positive

caregiver-child interactions.

• Interventions that focus on

infant and young child

feeding or exclusive

breastfeeding, without an

emphasis on responsiveness

between caregiver and

child.

• Interventions that

exclusively target caregivers

(e.g. through provision of

information or education),

rather than targeting the

caregiver-child dyad to

facilitate and encourage

quality caregiver-child

interactions.

• Interventions that focus

exclusively on the child.

Interventions that

only implement early

learning and

development

Interventions that enhance caregivers’ access,

knowledge, attitudes, practices, or skills with respect

to supporting early learning and development for

young children. These interventions may either: a)

directly support caregivers in providing new early

learning opportunities for their children; or b) build

caregiver capacities more generally, providing

information and guidance around healthy

newborn/child development or a range of nurturing

care topics. Interventions may incorporate aspects of

responsive caregiving or behaviour management, but

the overall goals and activities of interventions to

support early learning are broader in scope.

Interventions may be supplemented by messages

about a variety of different caregiving topics but

must include messaging around early learning and

development. Intervention goals that relate to

caregiving, but are not clearly specified, will be also

categorized as general caregiving interventions.

Specific examples of caregiving interventions to

support early learning may include:

• Interventions to promote dialogic caregiver-child

book readings or book sharing.

• Interventions that provide learning and play

materials, such as book gifting or

developmentally-appropriate toys, to increase

opportunities for early learning.

• Interventions that promote general caregiving


supporting the needs of

caregivers and families, but

do not include a specific

objective to support

caregiving skills for

promoting early learning

and child development.


reproductive, maternal,

newborn and child health,

but do not include a specific

objective to support

caregiving skills for

promoting child

development.

• Interventions that are

specifically focused on

particular aspects of

caregiving (e.g. only

behaviour management,

only responsive caregiving)

20 | P a g e


Excluded

competencies to support early learning and

development in young children. These

interventions primarily focus on and support

caregivers themselves, as opposed to enhancing

the caregiver-child relationship. Examples

include caregiver group meetings to share

information and discuss caregiving issues;

home-visiting programmes to improve caregiver

knowledge of ECD and caregiving skills; or

informational sessions providing general advice

on caregiving covering discipline, routines,

feeding, and child health and development.

Interventions that

promote both

responsive

caregiving and the

promotion of early

learning/development

Caregiving interventions that combine

features/components of responsive caregiving and the

promotion of early learning (as defined above).

All caregiving

interventions

categorized above

Recognizing the variation of caregiving intervention

descriptions, this category captures all interventions

together categorized as either responsive caregiving,

early learning promotion, or a combination of

responsive caregiving and the promotion of early

learning.

Caregiving to

support healthy

socioemotional and

behavioural

development

Interventions that support caregivers in promoting

healthy socioemotional and behavioural development

for young children and preventing child behaviour

problems or child maltreatment. This includes

encouraging caregivers to use appropriate and

desirable practices, including sensitive discipline and

limit-setting; reduce inappropriate behaviour

management practices, such as harsh discipline and

coercion; or some combination. Positive parenting

and behaviour management interventions encourage

stable and healthy family relationships and provide

for the physical and emotional safety of the child to

promote positive behavioural development outcomes

for young children. Examples of interventions for

caregiving to support healthy child socioemotional

and behavioral development include:

• Interventions promoting positive behaviour

management techniques, such as establishing

daily routines, praise and appropriate discipline.

• Interventions reducing child maltreatment and

associated factors, such as harsh punishment.

• Interventions that relate to

caregiving more generally,

in which behaviour

management or child

behavioural development is

not the primary focus.

• Interventions intended to

promote early learning

opportunities.

• Interventions where the

primary goal is secure

attachment and supportive,

sensitive and responsive

interactions between

caregivers and children

more broadly.

Combined caregiving

and nutrition

interventions

Interventions that combine a caregiving component

with a nutrition component such as:

• Caregiving component: interventions that

enhance caregivers’ access, knowledge,

attitudes, practices, or skills with respect to

supporting caregiving (responsive caregiving,

caregiving to support early learning, healthy

socioemotional and behavioural development for

• Interventions that contain

only caregiving components

or only nutrition

components.

• Interventions that do not

assess a child development

outcome.

21 | P a g e


Excluded

young children).

• Nutrition component: may include breast

feeding promotion, agricultural or nutrition

education or provision of a macronutrient or

micronutrient supplement.

• Interventions that promote

nutrition through agriculture

only (e.g. livestock, crops).

• Interventions that promote

water, sanitation and

hygiene only.

22 | P a g e

Evidence and recommendations for responsive caregiving interventions (PICO 1)

Review question

What are the effects of responsive caregiving interventions on ECD in the first 3 years of life?

P – Caregivers and their children in the first 3 years of life

I – Responsive caregiving interventions [alone]

C – Standard of care or comparison groups without responsive caregiving interventions

O – ECD (primary), child attachment, child growth, child health and nutrition, caregiving

knowledge, caregiving practices, caregiver-child interactions, and caregivers’ mental health.

Summary of evidence

The GRADE table and forest plots for the meta-analyses are shown in Appendix D. We

identified 17 responsive caregiving programmes for caregivers and their children during the first

3 years of life. The majority (n=13) of programmes were conducted in HICs, specifically

Australia (Wake et al., 2011), Canada (Barrera et al., 1986), Japan (Cheng et al., 2007), the

Netherlands (Juffer et al., 1997; Van Zeijl et al., 2006; Velderman et al., 2006), the United

Kingdom (McGillion et al., 2017), and the United States (Dozier et al., 2009; Guttentag et al.,

2014; Kochanska et al., 2013; Mendelsohn et al., 2007; Spieker et al., 2012; Weisleder et al.,

2016). The four remaining programmes were conducted in Bangladesh (Frongillo et al., 2017),

Chile (Santelices et al., 2011), Lithuania (Kalinauskiene et al., 2009), and South Africa (Cooper

et al., 2009; Murray et al., 2016).

Three programmes used a cluster randomized control trial (c-RCT) study design, and 14

used a randomized control trial (RCT) design with randomization at the individual or household

level. Studies ranged in publication date from 1986 to 2017. The programmes evaluated between

43 (Barrera et al., 1986) and 4365 individuals (Frongillo et al., 2017). They varied in the

populations that they targeted at enrollment. Three programmes enrolled expectant mothers

(Cooper et al., 2009; Guttentag et al., 2014; Santelices et al., 2010), three targeted caregivers and

their newborn infants during the first 3 months of life (Weisleder et al., 2016; Juffer et al., 1997;

Mendelsohn et al., 2007), five focused on infants during the first year of life (Barrera et al., 1986;

Kalinauskiene et al., 2009; McGillion et al., 2017; Velderman et al., 2006; Cheng et al., 2007),

and six were conducted with a broader age range of children, from birth to 4 years old (Frongillo

et al., 2017; Spieker et al., 2012; Kochanska et al., 2013; Van Zeijl et al., 2006; Wake et al.,

2011; Dozier et al., 2009). All programmes focused on engaging mothers and their children.

The programmes also varied in terms of implementation. Dosage and duration ranged from

a shorter intervention delivered in weekly sessions over 2.5 months (Spieker et al., 2012) to a

longer intervention delivered over three years (Weisleder et al., 2016). Twelve programmes were

delivered through individualized home visits, two were delivered through individualized sessions

at a research center or clinic, one was delivered through caregiver groups, and two used a

combination of both individualized visits and caregiver groups.

23 | P a g e

Early child outcomes:

The effects of responsive caregiving interventions on early child outcomes are categorized and

presented in terms of the following outcome domains: cognitive development, language

development, motor development, socioemotional development, attachment, height-for-age

(HAZ), and weight-for-age (WAZ).

Cognitive development: Three studies assessed programme impact on cognitive development.

All three of these used the Bayley Scale of Infant Development (BSID) (Murray et al., 2016;

Mendelsohn et al., 2007; Barrera et al., 1986). Only Mendelsohn and colleagues (2007)

presented the unadjusted means and standard deviations which could be extracted for this

analysis. Results from this study indicated that the impact on cognitive development was null

(SMD = 0.26, 95% CI: -0.14, 0.66; n = 1). The overall quality of evidence was graded as low.

Language development: Five programmes evaluated intervention impact on language

development. Two evaluations used the Landry Parent-Child Interaction Scales and Preschool

Language Scale (Guttentag et al., 2014; Mendelsohn et al., 2007), one used a measure developed

by the investigators (Frongillo et al., 2017), and two used a version of the MacArthur-Bates

Communicative Development Inventory (McGillion et al., 2017; Wake et al., 2011). The pooled

results showed no significant impacts on language development (SMD = 0.08, 95% CI: -0.07,

0.23; n = 5). The overall quality of evidence was graded as moderate.

Motor development: Two programmes evaluated the impact on motor development. One study

used the BSID (Barrera et al., 1992), and the other used a caregiver-reported motor milestone

checklist previously adapted by the authors (Frongillo et al., 2017). Frongillo and colleagues

(2017), was the only study to present unadjusted means and standard deviations to calculate the

effect size, and the results indicated a significant improvement in motor development (SMD =

0.19, 95% CI: 0.12, 0.26; n = 1). The overall quality of evidence was graded as moderate.

Socioemotional development: Four programmes evaluated the impact on socioemotional

development -- measured using the Infant–Toddler Social and Emotional Assessment (ITSEA)

(Guttentag et al., 2014; Kochanska et al., 2013; and Spieker et al., 2012) and the Behavior

Assessment System for Children (Weisleder et al., 2016). The pooled results showed no

significant effect on improving socioemotional development (SMD = 0.14, 95% CI: -0.03, 0.30;

n = 4). The overall quality of evidence was graded as low.

Behaviour problems: Seven programmes evaluated the impact on behaviour problems measured

using the Child Behavior Checklist (CBCL) (Cheng et al., 2007; Mendelsohn et al., 2007;

Spieker et al., 2014; Van Zeijl et al., 2006; Wake et al., 2011), the ITSEA (Guttentag et al.,

2014), and the BASC (Weisleder et al., 2016). The pooled results showed no significant effect on

reducing behaviour problems (SMD = -0.14, 95% CI: -0.29, 0.002; n = 7). The overall quality of

evidence was graded as low.

Attachment: Seven studies evaluated the impact of responsive caregiving interventions on

attachment outcomes. The measure that was most commonly used to assess attachment outcomes

was the Ainsworth Strange Situation procedure (Santelices et al., 2010; Juffer et al., 1997;

24 | P a g e

Velderman et al., 2006). Pooled results indicated no impacts on attachment outcomes (SMD =

0.13, 95% CI: -0.11, 0.37; n = 3). Three studies which could not be included in the pooled results

similarly reported null effects; one reported significant improvements in attachment outcomes

(Cooper et al., 2009). The overall quality of evidence was graded as low.

HAZ and WAZ: One programme evaluated the impact on HAZ and WAZ (Frongillo et al., 2017).

This study found a positive effect on improving HAZ (SMD = 0.10, 95% CI: 0.03, 0.16; n = 1).

The overall quality of evidence on HAZ was graded as moderate. This study found no significant

effect on improving WAZ (SMD = 0.03, 95% CI: -0.04, 0.10; n = 1). The overall quality of

evidence on WAZ was graded as moderate.

Caregiving outcomes:

The effects of responsive caregiving interventions on caregiving/parenting outcomes are

categorized and presented in terms of the following aspects of parenting: caregiving knowledge,

caregiving practices, caregiver-child interactions, and caregivers’ mental health.

Caregiving knowledge: One programme evaluated the impacts on caregiving knowledge

(Spieker et al., 2012). This study used Raising a Baby (RAB), a measure of caregiver knowledge

of infant and toddler socioemotional needs and developmentally-appropriate expectations. The

results showed no impact on caregiving knowledge (SMD = 0.29, 95% CI: -0.01, 0.58; n=1). The

overall quality of evidence was graded as low.

Caregiving practices: Three programmes evaluated the impacts on caregiving practices. One

evaluation measured caregiving practices using Home Observation for Measurement of the

Environment (HOME) (Barrera et al., 1986), one used a coded video observation measure

(Murray et al., 2016), and one study assessed parenting practices using the StimQ. The pooled

results showed no impacts on caregiving practices (SMD = 0.53, 95% CI: -0.10, 1.17; n = 2).

The overall quality of evidence was graded as low.

Caregiver-child interactions: Eight programmes evaluated impacts on caregiver-child

interactions. Caregiver-child interactions were observed and coded using measures including the

Landry Parent-Child Interaction Scales (Guttentag et al., 2014), the Nursing Child Assessment

Teaching Scale (Spieker et al., 2012), the Ainsworth Scales (Kalinauskiene et al., 2009;

Velderman et al., 2006; Juffer et al., 1997), and the Parent/Caregiver Involvement Scale (Cooper

et al., 2009). The pooled results showed a significant improvement in the quality of caregiver-

child interactions (SMD = 0.34, 95% CI: 0.15, 0.54; n=6). The overall quality of evidence was

graded as low.

Caregiver depressive symptoms: Three programmes evaluated the impacts on caregiver

depressive symptoms. One study used the Edinburgh Postnatal Depression Scale (Cooper et al.,

2009), one used the Beck Depression Inventory (Kalinauskiene et al., 2009), and the third used

the Center for Epidemiological Studies Depression Scale (CES-D) (Mendelsohn et al., 2007).

The pooled results showed that interventions significantly reduced caregiver depressive

25 | P a g e

symptoms (SMD = -0.21, 95% CI: -0.39, -0.04, n = 3). The overall quality of evidence was

graded as moderate.

Subgroup analyses

Subgroup analyses were conducted to examine possible moderating effects by HIC versus LMIC

for outcomes for which it was feasible. Results are presented in Table 5; it is noteworthy that

only two studies from LMICs had quantitative data available for the analysis.

Table 5. Child and caregiver outcomes for interventions that implement responsive caregiving only, by HICs

versus LMICs

Outcome

HICs LMICs

SMD 95% CI N SMD 95% CI N

Child outcomes

Language development 0.00 -0.15, 0.15 4 0.23 0.16, 0.30 1

Attachment 0.16 -0.10, 0.43 2 0.00 -0.53, 0.53 1

Caregiver outcomes

Caregiving practices 0.21 -0.19, 0.61 1 0.86 0.45, 1.28 1

Caregiver-child interactions 0.32 0.07, 0.58 4 0.46 -0.06, 1.00 2

Caregiver depressive symptoms -0.17 -0.56, 0.23 1 -0.22 -0.42, -0.03 2

Considerations for adverse effects and costs

There were no reported risks of adverse outcomes with responsive caregiving

interventions for ECD or parenting outcomes. None of the studies presented data on costs.

In conclusion, the global evidence suggests that responsive caregiving interventions during the

first 3 years of life are effective in improving early child motor development and caregiving

outcomes, specifically caregiver and child interactions and maternal depressive symptoms. More

studies are required from LMICs.

26 | P a g e

Evidence and recommendations for caregiving interventions that promote early

learning (PICO 2)

Review question

What is the effectiveness of caregiving interventions that promote early learning on ECD in the



I – Interventions that promote early learning [alone]

C – Standard of care or comparison groups without caregiving interventions to support early

learning


knowledge, caregiving practices, caregiver-child interactions, and caregivers’ mental health

Summary of evidence

GRADE table and forest plots for the meta-analyses are shown in Appendix E. The

evidence for caregiving interventions to promote early learning is derived from 22 RCTs. Four

trials were conducted in LMICs, including China, India, Uganda, and Zambia. Eighteen were

conducted in HICs, including Australia, Bermuda, France, Hong Kong, New Zealand, , the

United States and the United Kingdom (see Appendix C for full list of PICO 2 programmes and

characteristics). The majority were RCTs randomized at the individual level (n = 17), and five

studies were c-RCTs. Studies were published between the years 1988 and 2017. The analytic

sample sizes at endline ranged from 48 (Wasik et al., 1990) to 2557 participants (Goodson et al.,

2000).

Seven of the trials targeted women beginning during pregnancy (Goodson et al., 2000;

Guedeney et al., 2013; Jacobs et al., 2016; Love et al., 2005; Norr et al, 2003; Robling et al.,

2016; Walkup et al., 2009); four interventions targeted women immediately following childbirth

(Caughy et al., 2004; Schwarz et al., 2012; Brooks-Gunn et al., 1992; Nair et al., 2009); three

interventions enrolled women and children within the first 3 months of life (Fergusson et al.,

2005; Sawyer et al., 2017; Wasik et al., 1990); four targeted children within the first year of life

(Jin et al., 2007; Wagner et al., 2002; Rockers et al., 2016; Muhoozi et al., 2017); and four

enrolled children within their first 36 months (Hutchings et al., 2016; Leung et al., 2017b; Scarr

& McCartney, 1988; Heubner et al., 2000). Most programmes targeted the mother or other

female primary caregiver.

In terms of delivery, ten of the programmes were delivered to families individually

through home visitations, five were group-based, and seven included a mix of individual and

group-based delivery strategies. Programme duration varied extensively, ranging from 1.5

(Heubner et al., 2000) to 60 months (Goodson et al., 2000).

27 | P a g e


The effects of caregiving interventions to support early learning on early child outcomes

are categorized and presented in terms of the following outcome domains: cognitive

development, language development, motor development, socioemotional development, child

behaviour problems, attachment, HAZ and WAZ.

Cognitive development: Thirteen studies assessed programme impact on cognitive development.

Four of these used the BSID. The remaining ones used a variety of measures, including:

Wechsler Preschool and Primary Scale of Intelligence (Schwarz et al., 2012), Kaufman

Assessment Battery for Children (Goodson et al., 2000), INTERGROWTH-21st NDA tool

(Rockers et al., 2016), Preschool Developmental Assessment Scale (Leung et al., 2017b), an

author-created measure of standardized checklist items (Robling et al., 2016), McCarthy Scales

of Children’s Abilities (Wasik et al., 1990), Schedule of Growing Skills II (SGS-II) (Hutchings

et al., 2016), Stanford-Binet Test of Intelligence (Scarr & McCartney, 1988), and Developmental

Profile II (DPII) (Wagner et al., 2002). The pooled results indicated positive impacts of

caregiving interventions on cognitive development (SMD = 0.20, 95% CI: 0.01, 0.39; n = 8). The


Language development: Impacts on language development were assessed in nine studies. Three

used the Peabody Picture Vocabulary Test-Revised (Goodson et al., 2000; Heubner, 2000; Love

et al., 2005). Other measures included the Gesell Development Schedules (Jin et al., 2007), the

Stanford-Binet Test of Intelligence (Brooks-Gunn et al., 1992), the BSID (Muhoozi et al., 2017),

the Wechsler Preschool and Primary Scale of Intelligence (Schwarz et al., 2012), an Early

Language Milestone measure (Robling et al., 2016), and the Developmental Profile II (Wagner et

al., 2000). Pooled results indicated that interventions had no significant impacts on language

development (SMD = 0.07, 95% CI: -0.11, 0.24; n = 6). The overall quality of the evidence was

graded as low.

Motor development: Seven studies evaluated programme effects on motor development. Three

used the BSID (Muhoozi et al., 2017; Norr et al., 2003; Nair et al., 2009). Two studies used a

caregiver-reported measure (Wagner et al., 2002; Rockers et al., 2016), and two used direct

observation measures (Brooks-Gunn et al., 1992; Jin et al., 2007). The pooled results show

significant positive effects on motor development (SMD = 0.32, 95% CI: 0.12, 0.52; n = 5). The

overall quality of the evidence was graded as low.

Socioemotional development: Nine studies assessed programme impacts on socioemotional

development -- measured using the Ages and Stages Questionnaire (Muhoozi et al., 2017;

Pontoppidan et al., 2016; and Sawyer et al., 2017), the Adaptive Social Behavior Inventory

(Wagner et al., 2002), ITSEA (Barlow et al., 2007; Walkup et al., 2009), the Strengths and

Difficulties Questionnaire (SDQ) (Leung et al., 2017b), the Gesell Development Schedules (Jin

et al., 2007), and a measure from the NICHD Study of Early Child Care (Love et al., 2005).

Pooled results showed positive effects on socioemotional development outcomes (SMD = 0.28,

95% CI: 0.09, 0.48, n = 3). However, there were non-significant differences in five out of the six

other studies that could not be meta-analysed. The overall quality of the evidence was graded as

very low.

28 | P a g e

Behaviour problems: Eight programmes evaluated the impact on behaviour problems - measured

using the ITSEA (Fergusson et al., 2005; Jacobs et al., 2016; Walkup et al., 2009), the CBCL

(Caughy et al., 2004; Goodson et al., 2000; Love et al., 2005), the SDQ (Leung et al., 2017b),

and the PSI-DC (Hutchings et al., 2016). The pooled results showed no significant effect on

reducing behaviour problems (SMD = -0.25, 95% CI: -0.54, 0.04, n = 3). Of the five studies that

could not be meta-analysed, the evidence was mixed: three studies found reductions in children’s

behaviour problems, and two found no significant differences. The overall quality of evidence

was graded as very low.

Attachment outcomes: Two studies assessed programme impacts on attachment outcomes

(Caughy et al., 2004; Guedeney et al., 2013). Only one study (Guedeney et al., 2013) contributed

to the effect size estimate (measured reductions in social withdrawal using the Alarm Distress

Baby Scale; direction of effect size was reversed for analysis). Results indicated significant

positive impacts on attachment outcomes (SMD = 0.30, 95% CI: 0.09, 0.51; n = 1). Caughy and

colleagues (2004), similarly reported significant improvements in secure attachment

relationships for intervention group children (as measured by the Attachment Q-Set). The overall

quality of evidence was graded as low.

HAZ and WAZ: Two studies evaluated impacts on HAZ and WAZ (Muhoozi et al., 2017;

Rockers et al., 2016). The pooled results showed no effects on child HAZ outcomes (SMD = -

0.02, 95% CI: -0.29, 0.24, n = 2). The overall quality of the evidence was graded as moderate.

The pooled results showed no effects on child WAZ outcomes (SMD = 0.05, 95% CI: -0.10,

0.19, n = 2). The overall quality of the evidence was graded as moderate.

Caregiving outcomes

The effects of caregiving interventions to support early learning on caregiving outcomes are

categorized and presented in terms of the following aspects of parenting: caregiving knowledge,

caregiving practices, caregiver-child interactions, and caregivers’ mental health.

Caregiving knowledge: Three studies assessed intervention impacts on caregiving knowledge.

Two studies used questionnaires created by the authors for the purposes of the study (Jin et al.,

2007; Walkup et al., 2009), and one used the Knowledge of Infant Development Inventory

(Wagner et al., 2002). Two of the evaluations reported significant improvements in caregiving

knowledge (Jin et al., 2007; Walkup et al., 2009), while one study found no programme effects

(Wagner et al., 2002). Unadjusted means and standard deviations were not presented in the

papers, and therefore it was not possible to calculate a pooled estimate. The overall quality of the


Caregiving practices: Programme impacts on caregiving practices were assessed in eight studies.

Only one of the trials found significant improvements in caregiving practices (Love et al., 2005),

while the remaining studies reported null effects (Wasik et al., 1990; Walkup et al., 2009;

Wagner et al., 2002; Norr et al., 2003; Hutchings et al., 2016; Goodson et al., 2000; Caughy et

al., 2004). All of the studies used the HOME. Pooled results indicated no programme impacts on

29 | P a g e

caregiving practices (SMD = 0.05, 95% CI: -0.04, 0.13; n = 2). The overall quality of evidence

was graded as low.

Caregiver-child interactions: Five studies evaluated intervention impacts on caregiver-child

interactions. The majority of studies measured interactions using the Nursing Child Assessment

Satellite Training (NCAST) measure (Caughy et al., 2004; Goodson et al., 2000; Wagner et al.,

2002). Three studies reported positive impacts on caregiver-child interactions (Caughy et al.,

2004; Love et al., 2005; Hutchings et al., 2016), while two studies reported no impacts (Goodson

et al., 2000; Wagner et al., 2002). It was not possible to calculate a pooled estimate as studies did

not present the unadjusted means and standard deviations. Overall quality of evidence was rated

as low.

Caregiver depressive symptoms: Four studies assessed programme impacts on caregiver

depressive symptoms. Two studies used the Beck Depression Inventory (Schwarz et al., 2012;

Hutchings et al., 2016); one study used the WHO Self-Reporting Questionnaire (SRQ) z-score

(Rockers et al., 2016); and the fourth used the CES-D (Walkup et al., 2009). The pooled results

showed no effect on caregiver depressive symptoms (SMD = 0.07, 95% CI: -0.08, 0.22, n = 2).

The overall quality of the evidence was graded as moderate.

Subgroup analyses

Subgroup analyses were conducted to examine possible moderating effects by HICs

versus LMICs. Results are presented in Table 6. The impact on ECD and caregiver mental

health appears greater in LMICs; however, more studies measuring this outcome in both HICs

and LMICs are required, as currently we draw findings from only one study in each context.

Table 6. Child and caregiver outcomes for interventions that promote early learning and development, by

HICs versus LMICs

Outcome HICs LMICs


Child outcomes

Cognitive development 0.08 -0.02, 0.18 5 0.32 -0.11, 0.75 3

Language development 0.03 -0.09, 0.15 4 0.30 -0.67, 1.27 2

Motor development 0.08 -0.10, 0.26 1 0.39 0.17, 0.60 4

Caregiving outcomes

Caregiver depressive symptoms 0.13 -0.11, 0.37 1 0.03 -0.16, 0.22 1

30 | P a g e

Considerations for adverse effects and costs:

There were no indications of adverse effects caused by caregiving interventions to

support development and early learning. Costing data for these interventions were not readily

available in the literature.

In conclusion, the intervention studies on caregiving to support early learning suggest

these are promising interventions with significant (modest) effects found on child cognition,

motor development and attachment. Evidence for effectiveness on caregiving outcomes was not

observed in the pooled data. However, more studies from LMICs and more data on caregiver-

level outcomes are needed.

31 | P a g e

Evidence and recommendations for caregiving interventions that combine

responsive caregiving and promotion of early learning: additional analysis 2.1

Review Question

What is the effectiveness of caregiving interventions that combine both responsive caregiving

and the promotion of early learning on ECD in the first 3 years of life?


I – Caregiving interventions that combine responsive caregiving and the promotion of early

learning

C – Standard of care or comparison groups without caregiving interventions to support

responsive caregiving and the promotion of early learning



Summary of evidence

The GRADE table and forest plots for the meta-analyses are shown in Appendix F. We

identified 42 caregiving interventions that had components of both responsive caregiving and the

promotion of early learning for caregivers and their children during the first 3 years of life.

Approximately half (n=22 or 52.4%) of these programmes were conducted in LMICs. A total of

14 programmes used a c-RCT study design, and 28 programmes used a RCT design with

randomization at the individual level. Studies ranged in publication date from 1974 (Johnson et

al., 1974) to 2017 (Fernald et al., 2017; Helmizar et al., 2017; Leung et al., 2017a). The

programmes evaluated between 32 (Johnson et al., 1974; Pontoppidan et al., 2016; Whitt &

Casey, 1982) and 1411 individuals (Yousafzai et al., 2014). These programmes varied in targeted

population at enrollment. Seven programmes (16.7%) focused on expectant mothers beginning

during pregnancy, 13 rogrammes (31.0%) targeted caregivers and their newborn infants during

the first 3 months of life, six (14.3%) focused on infants during the first year of life, and the

remaining 16 (38.1%) were conducted either after the first year or among a broader age range of

children from birth to 4 years old. Nearly all programmes (n=38 or 90%) targeted mothers and

their children. Four notable programmes (10%) engaged both mothers and fathers (Johnson et al.,

1974; Kaaresen et al., 2008; Kyno et al., 2012; Singla et al., 2015).

The programmes also varied in terms of implementation. Dosage and duration ranged from

as short as fewer than 10 sessions over two months (Aboud & Ahkter, 2011; Vally et al., 2015;

Murray et al., 2016) to as long as weekly (Kaminski et al., 2013) or biweekly sessions

(Wallander et al., 2014) over three years. A total of 25 (60%) were individualized programmes,

six (14%) were delivered through caregiver groups, and 11 (26%) used a combination of both

home visits and caregiver groups.

32 | P a g e


Cognitive development: Thirty-six studies assessed programme impact on cognitive

development. Twenty-one of these used the BSID. Six studies used the Griffith’s scales of

Mental Development (Waber et al., 1981; Walker et al., 2004; Powell & Grantham-McGregor,

1989; Powell et al., 2004; Gardner et al., 2005; Grantham-McGregor et al., 1991). The remaining

studies utilized a variety of measures, including the Kaufman Assessment Battery for Children

(Drotar et al., 2009), Preschool Developmental Assessment Scale (PDAS) (Leung et al., 2017a),

McCarthy Scales of Children’s abilities (Fernald et al., 2017), a nationally validated author-

created checklist measure (Hartinger et al., 2017), Stanford-Binet Test of Intelligence (Johnson

et al., 1974; Madden et al., 1984), Cattell Scales (Olds et al., 1986), Bracken Basic Concept

Scale (Cronan et al., 1996), and Mullen Scale of Early Learning (Kyno et al., 2012). The pooled

results showed positive impacts of caregiving interventions on cognitive development (SMD =

0.45, 95% CI: 0.25, 0.65; n = 20). The overall quality of evidence was graded as low.

Language development: Seventeen studies assessed the effects on language development. Six

used the BSID. Four used the Griffiths Mental Development Scales (Chang et al., 2015; Gardner

et al., 2005; Grantham-McGregor et al., 1991; Powell et al., 2004). The MacArthur-Bates

Communicative Development Inventory was used in three studies (Vally et al., 2015; Goldfeld et

al., 2011; Cronan et al., 1996). Remaining studies used other measures of direct observation. The

pooled results indicated positive impacts on language outcomes (SMD = 0.38, 95% CI: 0.16,

0.60; n = 14). The overall quality of evidence was graded as low.

Motor development: Eighteen studies assessed programme impacts on motor development.

Twelve measured motor development using the BSID (Attanasio et al., 2014; Field et al., 1982;

Hamadani et al., 2006; Heinicke et al., 1999; Helmizar et al., 2017; Kaaresen et al., 2008; Lozoff

et al., 2010; Nahar et al., 2012a; Tofail et al., 2013; Vazir et al., 2013; Wallander et al., 2014;

Yousafzai et al., 2014). Five studies used the Griffith Mental Development Scale (Chang et al.,

2015; Gardner et al., 2005; Grantham-McGregor et al., 1991; Grantham-McGregor et al., 1989;

Powell et al., 2004), and one study used the Mullen Scales of Early Learning (Kyno et al., 2012).

Pooled results indicated significant positive effects on motor development outcomes (SMD =

0.25, 95% CI: 0.09, 0.40; n = 13). The overall quality of evidence was rated as low.

Socioemotional development: Four studies assessed programme impacts on socioemotional

development -- measured using the ASQ (Kyno et al., 2012), the SDQ (Kaminski et al., 2013),

the Social Skills Rating System (Drotar et al., 2008), and the BSID (Yousafzai et al., 2014).

Pooled results showed null effects on socioemotional development outcomes (SMD= 0.06, 95%

CI: -0.18, 0.28, n=2). The studies that could not be meta-analysed similarly found no statistically

significant differences for socioemotional development. The overall quality of the evidence was

graded as moderate.

Behaviour problems: Seven programmes evaluated the impact on behaviour problems measured

using the CBCL (Constantino et al., 2001; Kaaresen et al., 2008; Kitzman et al., 1997; Kyno et

al., 2012; Olds et al., 2002), the Devereux Early Childhood Assessment (Kaminski et al., 2013),

and the Eyberg Child Behavior Inventory (Leung et al., 2017a). The pooled results showed no

significant effect on reducing behaviour problems (SMD = -0.18, 95% CI: -0.40, 0.04, n = 2). Of

33 | P a g e

the five studies that could not be meta-analysed, four found no significant differences. The


Attachment outcomes: Two studies evaluated programme impacts on attachment outcomes.

Measures included the Ainsworth Strange Situation procedure (Heinicke et al., 1999) and the

Attachment Q-Set (Roggman et al., 2009). Both studies reported significant positive impacts on

attachment outcomes, although a pooled effect size could not be calculated. Overall quality of

evidence was graded as moderate.

HAZ and WAZ: Eight studies evaluated impacts on HAZ and WAZ. Two other studies assessed

height and weight but did not present a z-score; therefore, they were excluded from the meta-

analysis (Attanasio et al., 2014; Nair et al., 2009). The pooled results showed no effects on child

HAZ outcomes (SMD = -0.04, 95% CI: -0.15, 0.07, n = 8). The overall quality of the evidence

was graded as moderate. The pooled results showed no effects on child WAZ outcomes (SMD =

0.02, 95% CI: -0.07, 0.11, n = 6). The overall quality of the evidence was graded as high.


Caregiving knowledge: Seven studies examined programme impact on caregiving knowledge.

Four studies used caregiver self-report questionnaires developed by the authors (Hamadani et al.,

2006; Vazir et al., 2013; Powell et al., 2004; Singla et al., 2015). Pooled results indicated that

programmes had a significant positive impact on caregiving knowledge (SMD = 0.73, 95% CI:

0.57, 0.89; n = 6). The overall quality of evidence was rated as low.

Caregiving Practices: Eighteen studies assessed intervention effects on caregiving practices. The

majority of studies used the HOME (Aboud & Akhter, 2011; Aboud et al., 2013; Barlow et al.,

2007; Chang et al., 2015; Heinicke et al., 1999; Helmizar et al., 2017; Johnson et al., 1974;

Kitzman et al., 1997; Nahar et al., 2012b; Singla et al., 2015; Walker et al., 2004; Yousafzai et

al., 2015; Olds et al., 1986; Field et al., 1982). Two studies used the Family Care Indicators

(Attanasio et al., 2014; Tofail et al., 2013). Pooled results indicated that interventions had a

significant positive effect on caregiving practices (SDM = 0.48, 95% CI: 0.20, 0.76; n = 10). The

overall quality of evidence was rated as low.

Caregiver-child interactions: Programme impacts on caregiver-child interactions were assessed

in 12 studies. Measures to assess the quality of mother-child interactions included the NCAST

(Kitzman et al., 1997) and the Observation of Mother and Child Interaction (OMCI) (Yousafzai

et al., 2015). Other measures of caregiver-child interactions included assessments of responsive

talk (Aboud & Akhter, 2011) and reciprocity during book-sharing activities (Murray et al.,

2016). The pooled results indicated that interventions had a significant positive effect on

caregiver-child interactions (SMD = 0.74, 95% CI: 0.39, 1.10; n = 5). Overall quality of evidence

was rated as moderate.

Caregiver depressive symptoms: Nine studies assessed programme impacts on caregiver

depressive symptoms. Six used the CES-D (Chang et al., 2015; Aboud et al., 2013; Singla et al.,

2015; Nahar et al., 2015; Attanasio et al., 2014; Baker-Henningham et al., 2005). Of the

remaining studies, one used the SRQ (Yousafzai et al., 2015), one used the Major Depression

34 | P a g e

Inventory (Pontoppidan et al., 2016), and one used a study-specific measure (Heinicke et al.,

1999). Pooled results indicated that interventions did not significantly impact caregiver

depressive symptoms (SMD = -0.08, 95% CI: -0.31, 0.15; n=7). The overall quality of the


Subgroup analyses:


versus LMICs. Results are presented in Table 7. The number of studies in LMICs is higher, and

a greater impact on outcomes is observed in these contexts, with the exception of caregiver-child

interactions.

Table 7. Child and caregiver outcomes for interventions that combine responsive caregiving and promotion

of early learning by HICs versus LMICs

Outcome HICs LMICs


Child outcomes

Cognitive development 0.23 -0.11, 0.58 4 0.49 0.27, 0.71 16

Language development 0.08 -0.24, 0.40 3 0.47 0.24, 0.70 11

Motor development -0.11 -0.60, 0.38 1 0.27 0.11, 0.42 12

Caregiving outcomes


Caregiver-child interactions 0.95 0.52, 1.38 2 0.64 0.16, 1.12 3

Caregiver depressive symptoms 0.40 -0.09, 0.89 1 -0.13 -0.37, 0.11 6


There appears to be no undue risk of adverse outcomes with interventions for combined

responsive caregiving and promotion of early learning. Few interventions evaluate cost. A cost-

effectiveness study conducted in Pakistan (Yousafzai et al., 2014) indicated US$ 48 per child per

year when delivered bundled with basic health and nutrition services (Gowani et al., 2014). In

settings where home visiting services or community groups do not already exist or where

resources are not being adequately allocated for ECD, such interventions are likely to increase

costs.

In situations where this type of intervention (e.g. psychosocial stimulation) is being integrated

within existing primary care and health services, care must be taken to ensure that there are no

adverse effects of adding the intervention on other child outcomes for health and nutrition,

35 | P a g e

particularly in low-resource contexts where health service capacities may be limited. A study

from the Caribbean explored these issues and found no negative impacts of integrating a

psychosocial stimulation intervention within existing health services on child nutrition or

immunization (Chang et al., 2015).

In conclusion, interventions that combine both features of caregiving (i.e., responsive

care and support for early learning) had significant positive effects for cognitive, language and

motor development, as well as caregiving knowledge, caregiver practices, and caregiver-child

interactions. Studies in 2.1 had global representation from both HICs and LMICs.

36 | P a g e

Evidence and recommendations for all caregiving interventions that deliver

responsive caregiving, promotion of early learning, or combined responsive

caregiving and promotion of early learning: additional analysis 2.2

Review question

What is the effectiveness of any caregiving interventions (responsive caregiving, promotion of

early learning, or combined responsive caregiving and the promotion of early learning) on ECD

in the first 3 years of life.


I – Any caregiving interventions (responsive caregiving, promotion of early learning or

combined responsive caregiving and the promotion of early learning)

C – Standard of care or comparison groups without caregiving interventions



Summary of evidence

The GRADE table and forest plots for the meta-analyses are shown in Appendix G. Here

we explore the effectiveness of caregiving interventions included in PICO questions 1, 2 and and

2.1. We describe the effectiveness of any caregiving interventions (responsive caregiving,

promotion of early learning, or combined responsive caregiving and the promotion of early

learning) on ECD in the first 3 years of life.


Cognitive development: Fifty-two studies assessed programme impact on cognitive development.

Over half of these used the BSID. Remaining studies used a variety of measures, the majority of

which were direct observation. The pooled results showed positive impacts of caregiving

interventions on cognitive development (SMD = 0.37, 95% CI: 0.22, 0.52, n = 29). The overall


Language development: Overall, 31 studies evaluated intervention effects on language

development. The pooled results indicated positive effects of caregiving interventions on

language outcomes (SMD = 0.24, 95% CI: 0.11, 0.36; n = 25). The overall quality of evidence

was graded as low.

37 | P a g e

Motor development: Twenty-seven studies assessed programme impacts on child motor

development outcomes. Pooled results indicated significant positive effects (SMD = 0.27, 95%

CI: 0.17, 0.37; n = 19). The overall quality of evidence was graded as moderate.

Socioemotional development: Seventeen studies assessed programme impacts on socioemotional

development. Pooled results indicated significant positive effects on socioemotional

development outcomes (SMD = 0.15, 95% CI: 0.04, 0.27, n = 9). The overall quality of the


Behavioural development: Twenty-two studies assessed programme impacts on behavioural

problems. Pooled results indicated significant reductions in behavioural problems (SMD = -0.17,

95% CI: -0.28, -0.06, n = 12). However, the majority of studies that could not be meta-analysed

found non-significant differences in behavioural problems. The overall quality of the evidence

was graded as low.

Attachment outcomes: Eleven studies evaluated the effects of caregiving interventions on

attachment outcomes. Pooled results indicated significant positive impacts of caregiving

interventions on attachment outcomes (SMD = 0.23, 95% CI: 0.07, 0.38; n = 4). The overall


HAZ and WAZ: Eleven studies evaluated impacts on HAZ and WAZ. The pooled results showed

no effects on child HAZ outcomes (SMD = -0.02, 95% CI: -0.10, 0.07, n = 11). The overall

quality of the evidence was graded as high. The pooled results showed no effects on child WAZ

outcomes (SMD = 0.03, 95% CI: -0.02, 0.08, n = 9). The overall quality of the evidence was

graded as high.

Child health outcomes: Four programmes evaluated the impact on child sickness, as reported by

the child’s primary caregiver. Three studies found no effects on reducing sickness (Aboud et al.,

2013; Menon et al., 2016; Singla et al., 2015), whereas another study found reductions in

diarrhoea and acute respiratory illness (Yousafzai et al., 2014).


Caregiving knowledge: Eleven studies measured the impact of interventions on caregiving

knowledge. Over half used a study-specific questionnaire developed by the authors. Pooled

results indicated that interventions had a significant positive effect on caregiving knowledge

(SMD = 0.68, 95% CI: 0.51, 0.85; n = 7). The overall quality of the evidence was graded as low.

Caregiving practices: Overall, 29 studies assessed programme impact on caregiving practices.

The majority of these (n = 23) measured caregiving practices using the HOME. Pooled results

indicated that interventions significantly improved caregiving practices (SMD = 0.44, 95% CI:

0.21, 0.67; n = 14). The overall quality of evidence was graded as low.

Caregiver-child interactions: Across all caregiving interventions, 25 studies assessed the effects

on caregiver-child interactions. All of these used a measure for coding video- or live-

observations of the mother-child dyad. The pooled results for intervention effects on interactions

38 | P a g e

indicated that programmes significantly improved outcomes for caregiver-child interactions

(SMD = 0.54, 95% CI: 0.30, 0.78; n = 11). The overall quality of the evidence was rated as low.

Caregiver depressive symptoms: Sixteen evaluations assessed programme impacts on caregiver

depressive symptoms. The majority of studies used either the CES-D or the Beck Depression

Inventory. The pooled results indicated that interventions did not significantly reduce caregiver

depressive symptoms (SMD = -0.07, 95% CI: -0.22, 0.07; n = 12). The overall quality of

evidence was rated as low.

Subgroup analyses:


versus LMICs. Results are presented in Table 8. In LMICs a greater impact on outcomes is

observed, with the exception of attachment.

Table 8. Child and caregiver outcomes across all included caregiving interventions by HICs versus LMICs

Outcomes HICs LMICs


Child outcomes

Cognitive development 0.12 0.01, 0.23 10 0.46 0.26, 0.65 19

Language development 0.02 -0.07, 0.11 11 0.42 0.23. 0.61 14

Motor development 0.06 -0.11, 0.23 2 0.29 0.19, 0.40 17

Attachment 0.25 0.08, 0.41 3 0.00 -0.53, 0.53 1

Caregiving outcomes

Caregiver knowledge 0.29 -0.01, 0.58 1 0.73 0.57, 0.89 6


Child-caregiver practices 0.48 0.20, 0.76 6 0.58 0.20, 0.95 5

Caregiver depressive symptoms 0.10 -0.16, 0.37 3 -0.12 -0.29, 0.05 9

In conclusion, benefits of caregiving interventions are observed on multiple outcomes of

ECD (with significant impacts on cognition, language, motor and behavioural development, and

attachment) as well as caregiving knowledge and practices, and caregiver and child interactions.

39 | P a g e

Evidence and recommendations for caregiving interventions to support healthy

child socioemotional and behavioural development (PICO 3)

Review question

What are the effects of caregiving interventions to support child socioemotional and behavioural

development on ECD?


I – Caregiving interventions that support socioemotional and behavioural development

C – Standard of care or comparison groups without caregiving interventions to support healthy

child socioemotional and behavioural development

O – ECD (primary), child growth, child health and nutrition, caregiving knowledge, caregiving

practices, caregiver-child interactions, and caregivers’ mental health.

Summary of Evidence6

The GRADE table and forest plots for the meta-analyses are shown in Appendix H. We

identified 10 caregiving studies to support healthy socioemotional development and behaviour

for children during the first 3 years of life. The evidence for the impact of caregiving

interventions to support socioemotional and behavioural development for children under 3 years

of age is derived from 10 RCTs representing 11 interventions. All 10 studies were conducted in

HICs, including Australia (n = 2), the Netherlands (n = 1), New Zealand (n = 1) and the United

States (n = 6). The majority were two-arm RCTs randomized at the individual level (n = 8), with

one two-armed c-RCT (Hiscock et al., 2008) and one three-armed c-RCT (Hiscock et al., 2018).

Studies were published between the years 2005 and 2018. The analytic sample sizes at endline

ranged from 237 (Van Zeijl et al., 2006) to 1353 (Hiscock et al., 2018). The median sample size

across all studies was 388 individuals.

Programmes varied in terms of the targeted caregivers. Most interventions (n = 7)

primarily targeted mothers or the primary caregiver of the child (Caldera et al. 2007; Gross et al.,

2009; Breitenstein et al., 2012; Fergusson et al., 2005; Hiscock et al., 2008; Hiscock et al., 2018;

Dishion et al., 2008). One programme restricted enrollment to children from dual-caregiver

families (i.e., biological mothers and father-figures) (Van Ziejl et al., 2006). Two programmes

specifically targeted adolescent and young mothers (Barlow et al., 2015; Jacobs et al. 2016).

Programmes also differed in the age of the child targeted for enrollment. Three

programmes targeted pregnant mothers, but the window for enrollment varied, through 32-weeks

gestation (Barlow et al., 2015), birth (Caldera et al., 2007), and the child’s first year of life

6 The terms ‘programmes’ and ‘studies’ are used interchangeably. The term ‘programme’ does not refer to routine service delivery in this report.

40 | P a g e

(Jacobs et al., 2016). One programme targeted children aged 0-3 months (Fergusson et al., 2005);

one enrolled children aged 1-3 years (Van Ziejl et al., 2006); one enrolled children aged 2-3

years (Dishion et al., 2008); two targeted caregivers with children aged 2-4 years (Gross et al.,

2009; Breitenstein et al., 2012); and three enrolled infants at their 8-month well-child visits

(Hiscock et al., 2008; Hiscock et al., 2018).

The majority of programmes were home visitation (n = 6) delivered to individual

families. Two studies evaluated group caregiver-training programmes affiliated with day-care

centres (Gross et al., 2009; Breitenstein et al., 2012). Two programmes contained both group and

individualized components, one of which was clinic-based only while the other also incorporated

in-home visitation (Hiscock et al., 2008; Hiscock et al., 2018).

There was a wide range in programme duration and dosage, from two-hour visits over

seven months (Hiscock et al., 2008) to home visiting interventions with up to 40 visits through

the child’s third birthday (Barlow et al., 2015). Median programme duration was 19.35 months.


The effects of interventions to support healthy socioemotional development and

behaviour on early child outcomes are categorized and presented in terms of the following

outcome domains: cognitive development, motor development, socioemotional development,

behavioural problems, and any other child nutrition or health outcome. None of the studies

assessed outcomes for language development or growth outcomes, and therefore these child

outcomes have also been excluded from the meta-analysis.

Cognitive development: Only one study assessed cognitive development. Cognition was

measured using the Mental Development Index score from the BSID. Mean scores for

intervention group children were significantly higher compared to children in the control group

(p<0.05) (Caldera et al., 2007). Unadjusted means and standard deviations were not presented so

the unadjusted effect size could not be calculated, although the authors indicate an adjusted

effect size of 0.29. The overall quality of evidence was graded as very low.

Motor development: One study assessed motor development (Caldera et al., 2007). Motor

development was measured using the Psychomotor Development Index score from the BSID.

Unadjusted effect sizes could not be calculated, but authors presented an adjusted effect size of

0.19, although the difference between groups was not statistically significant (p = 0.16). The

overall quality of evidence was graded as very low.

Socioemotional development: Barlow and colleagues (2015) was the only study to assess

socioemotional development. This was measured using the parental-reported Competence score

from the ITSEA. No significant effects were found (adjusted mean difference in intervention

41 | P a g e

group versus control scores: 0.04; p = 0.09; adjusted ES = 0.14). The overall quality of the

evidence was graded as very low.

Child behaviour problems: All 10 studies evaluated child behaviour problems. The pooled

results showed no effect on reductions in child behaviour problems (SMD = -0.02, 95% CI: -

0.07, 0.02, n = 5). Five studies used the parent-reported CBCL; two studies used the parent-

reported ITSEA (Barlow et al., 2015, Fergusson et al., 2005); one study used the Brief-ITSEA

(Jacobs et al., 2016); and two used the parent-reported Eyberg Child Behavior Inventory,

teacher-reported Caregiver-Teacher Report Form, and direct observation of child behaviour

problems assessed from videotaped play sessions (Gross et al., 2009; Breitenstein et al., 2012).

The overall quality of the evidence was graded as moderate.

Child health: Two studies examined programme impacts on indicators for child health and

medical outcomes. Outcomes were assessed through maternal interviews and review of medical

records. Caldera and colleagues (2007) indicated significant improvements in the number of

families with health care coverage for the child, but no effects on immunizations, receipt of well-

child visits, incidence of injuries requiring medical care, or number of hospitalizations and

emergency department visits. Fergusson and colleagues (2005) found intervention effects on

children being up-to-date on well-child visits, experiencing fewer hospitalizations for accidents

and injuries, and having higher rates of preschool dental services enrollment, but no effects were

observed on immunization rates. Child health outcomes were not commonly or consistently

measured across studies; therefore, child health was not included in the meta-analysis.


The effects of interventions to support healthy socioemotional development and

behaviour on caregiving outcomes are categorized and presented in terms of the following

aspects of parenting: caregiving knowledge, caregiving practices, caregiver-child interactions,

child maltreatment, caregivers’ mental health, and caregivers’ self-efficacy.

Caregiving knowledge: Two studies assessed programme impacts on caregiving knowledge.

Caldera and colleagues (2007) assessed maternal knowledge of infant development using the

Knowledge of Infant Development Inventory, and Barlow and colleagues (2015) assessed

caregiving knowledge through an author-designed maternal-reported questionnaire. Caldera and

colleagues (2007) found no significant effects on maternal knowledge, while Barlow and

colleagues (2015) observed significant improvements in caregiving knowledge among

intervention mothers. Unadjusted means and standard deviations were not presented, so pooled

results could not be calculated. The overall quality of evidence was graded as very low.

42 | P a g e

Caregiving practices: Eight evaluations examined programme impacts on caregiving practices.

Two studies used the HOME (Caldera et al., 2007; Barlow et al., 2015). Seven studies examined

parenting discipline strategies, two using the self-reported Parent Questionnaire (Breitenstein et

al., 2012; Gross et al., 2009), two using the self-reported Parent Behavior Checklist (Hiscock et

al., 2008; Hiscock et al., 2018), and three using the self-reported Parent-Child Conflict Tactics

Scale (Jacobs et al. 2016; Fergusson et al., 2005; Caldera et al., 2007). One study additionally

generated two parenting practices scores through factor analysis of the caregiver self-reported

Child Rearing Practices Report and the Adult-Adolescent Parenting Inventory measures

(Fergusson et al., 2005). The pooled results showed no effect on reductions in child behaviour

problems (SMD= 0.01, 95% CI: -0.04, 0.06, n=2). The overall quality of evidence was graded as

low.

Caregiver-child interactions: Five evaluations assessed programme effects on caregiver-child

interactions. One programme used the NCAST scale (Caldera et al., 2007). Two studies assessed

caregiver use of praise and commands during observed play and clean-up sessions, coded with

the Dyadic Parent-Child Interactive Coding System-Revised (Breitenstein et al., 2012; Gross et

al., 2009). Dishion and colleagues (2008) created a composite positive behaviour support score,

drawing items from HOME’s involvement subscale and direct observation measures. Van Ziejl

and colleagues (2006) assessed sensitivity, positive and negative discipline scores through three

direct observation measures during laboratory tasks. The pooled results showed no significant

effect on improving caregiver-child interactions (SMD = 0.14, 95% CI: -0.07, 0.34, n = 1). The

overall quality of evidence was graded as very low.

Child maltreatment: Two studies assessed programme impacts on child maltreatment. One study

examined parental report of contact with the Child, Youth and Family Service for issues relating

to child abuse and neglect (Fergusson et al., 2005) but did not see reductions in agency contacts.

The other (Jacobs et al., 2016) examined records from Child Protective Services to assess

whether substantiated child maltreatment reports had been filed, but also did not observe any

significant effects.

Caregiver mental health: Four studies assessed intervention impacts on caregiver mental health.

Two studies used the maternal self-reported CES-D (Barlow et al., 2015; Shaw et al., 2009) and

two studies used the primary caregiver-reported Depression Anxiety and Stress Scales (Hiscock

et al., 2008; Hiscock et al., 2018). The pooled results showed no effect on reductions in maternal

depressive symptoms (SMD = -0.05, 95% CI: -0.11, 0.01, n = 3). The overall quality of the


Self-efficacy: Three studies assessed intervention impacts on caregiver self-efficacy. Caldera and

colleagues (2007) measured maternal self-efficacy using the Teti Maternal Self-efficacy scale;

Breitenstein and colleagues (2012) and Gross and colleagues (2009) used the Toddler Care

Questionnaire. Both Caldera and colleagues (2007) and Breitenstein and colleagues (2012) found

43 | P a g e

intervention group mothers to have significantly higher self-reported self-efficacy scores

compared to control groups; however, Gross and colleagues (2009) found no effects. Unadjusted

means and standard deviations were not presented so pooled results could not be calculated. The


Subgroup analyses:

Subgroup analyses were conducted to examine possible moderating effects by intended

programme intensity (infrequent contacts or for less than 3 months versus regular contacts and

longer than 3 months) and programme delivery (individual versus group-based/mixed contacts).

None of the pooled effect sizes on child and caregiver outcomes differed by programme intensity

or programme delivery. It is likely that statistical power to detect differences by subgroup was

limited due to the small number of studies for many of the caregiver and child outcomes.


There were no indications of adverse effects caused by caregiving interventions to

support child socioemotional and behavioural development. Costing data were not readily

available. One study reported data on intervention costs (Hiscock et al., 2018). Mean costs of

this trial intervention programme were AUS$ 218 (AUS$ 208 costs to government and AUS$ 10

costs to family) and AUS$ 682 (AUS$ 516 costs to government and AUS$ 166 costs to family)

per family in the targeted and combined arms of the intervention respectively.

In conclusion, the global evidence on caregiving interventions to support children’s

socioemotional wellbeing and behaviour during the first three years of life is from HICs.

Therefore, these findings cannot be generalized to LMICs.

44 | P a g e

Evidence and recommendations for combined caregiving and nutrition

interventions (PICO 4)

Research questions

What are the effects of combined caregiving and nutrition programmes on ECD and child growth

outcomes in the first 3 years of life?

• What are the independent and additive effects of caregiving and nutrition interventions on

ECD and child growth outcomes in the first 3 years of life?

• Do the effects on ECD and child growth outcomes differ between programmes that are

targeted for young children with moderate to severe malnutrition compared to universal

programmes?


I – Combined caregiving (responsive caregiving, early learning promotion, or combined

responsive caregiving and early learning promotion) and nutrition programmes

C – a) standard of care; b) caregiving intervention alone; c) nutrition intervention alone



Summary of Evidence7

The GRADE table and forest plots for the meta-analyses are shown in Appendix I. We

identified 18 combined caregiving and nutrition interventions delivered to caregivers and their

young children during the first 3 years of life. Table 9 shows all 18 interventions (see Appendix

C for further details of programmes and characteristics). All 18 studies were conducted in

LMICs: Bangladesh (n = 6), Chile (n = 1), Colombia (n = 2), India (n = 1), Indonesia (n = 1),

Jamaica (n = 3), Pakistan (n = 1), Uganda (n = 2) and Zambia (n = 1). The RCTs varied in the

number of trial arms: 2-arms (n = 9), 3-arms (n = 2), 4-arms (n = 5), 5-arms (n = 1) and 6-arms

(n = 1). The majority of studies enrolled children from 6 months of age; however, Waber and

colleagues (1981) recruited caregivers during pregnancy, and Vazir and colleagues (2013) and

Yousafzai and colleagues (2014) enrolled children less than 6 months old.

With respect to the nutrition component of the intervention, nine studies provided

nutrition supplementation, typically with nutrition education, and the remainder focused on

nutrition education alone. Seven studies specifically targeted undernourished children (Gardner

7 The terms ‘programmes’ and ‘studies’ are used interchangeably. The term ‘programme’ does not refer to routine

service delivery in this report.

45 | P a g e

et al., 2005; Grantham-McGregor et al., 1991; Hamadani et al., 2006; Lozoff et al., 2010; Nahar

et al., 2012a; Powell et al., 2004; Tofail et al., 2013). With respect to the studies by Lozoff and

colleagues (2010) and Tofail and colleagues (2013) from Chile and Bangladesh respectively,

only the iron-deficient anemia arms of the trials were analysed. No studies were retrieved

combining caregiving and over-nutrition meeting the specified inclusion criteria in the initial

search.

Most interventions targeted mothers and children, with the exception of Singla and

colleagues (2015) where the targeting of mothers and fathers as caregivers was specified. The

interventions largely employed individual contacts or combined group and individual contacts,

with only two interventions using a group-only contact mode of delivery (Aboud & Akhter,

2011; Muhoozi et al., 2017). The average duration of implementation was 14 months, ranging

from two months (with a booster at six months) (Aboud & Akhter, 2011) to 36 months (Waber et

al., 1981).

Table 9. Overview of study design, intervention and comparison arms

Author Study design Combined intervention Caregiving intervention Nutrition intervention Standard of care

Aboud et al., 2013 2 arm Messages + illustrative card

provided around hygiene,

responsive feeding, play,

communication, gentle

discipline, and nutritious

foods/dietary diversity.

Addressed parenting

practices related to health,

nutrition, communication and

play.

Standard care - home visits

by government-paid family

welfare assistants with

messages about feeding and

hygiene

Aboud & Akhter,

2011

3-arm RCT The Responsive Feeding and

Stimulation (RFS) + group

received the RFS

intervention, plus Sprinkles

micronutrient powder for 6

months.

The RFS component of the

intervention comprised of 6

sessions that delivered 6

messages on "responsive

feeding and stimulation". In

the sessions, mothers could

participate in discussions and

practice sessions with their

children, while being

coached by the peer

educator. The peer educator

also demonstrated one

stimulation and responsive

feeding activity in each

session.

The control group received

12 sessions on health and

nutrition education and

information on child

development

Attanasio et al, 2014 4 arms (2x2

factorial)

Psychosocial stimulation and

micronutrient

supplementation

Psychosocial stimulation:

home visitors demonstrated

play activities using low-cost

or homemade toys, picture

books, and form boards.

These materials were left in

the homes for the week after

the visit and were changed

weekly.

The micronutrient

supplementation consisted of

Sprinkles delivered to

households every two weeks.

Standard of care

47 | P a g e


Frongillo et al.,

2017

Menon et al., 2016

2 arm Included intensive

interpersonal counselling on

infant and young child

feeding, responsive feeding,

mass media campaign, and

community mobilization

Alive and thrive usual

package (nutrition

counselling and mass media

campaign)

Gardner et al., 2005 4 arm (2X2

factorial

design; plus a

second non-

stunted

control

group, not

included in

this analysis)

Participants in the integrated

group received both

psychosocial stimulation and

zinc supplementation

Weekly, 30-minute

psychosocial stimulation

sessions were conducted by

trained community health

workers and focused on

improving maternal-child

interactions. In these

sessions, mothers learned

about activities to engage

with their child in an age-

appropriate fashion and

received simple toys from

the programme.

The nutrition component of

the intervention comprised

weekly zinc supplementation


placebo and routine care but

no stimulation

Grantham-

McGregor et al.,

1991

Walker et al., 1991

4 arm (2X2

factorial

design, plus

non-stunted

control

group)

Participants received both the

supplementation and

stimulation services

In the stimulation component

of the intervention, stunted

children received weekly

hour-long home visits from

community health aides that

focused on play and

stimulation. Mothers in this

group were instructed on

how to play with children

and impact their

development. Toys left

behind after the visit

facilitated mother-child play

and interactions in-between

two visits.

The supplementation

component of the

intervention comprised of

weekly distribution of 1kg of

milk-based formula for the

duration of the intervention.

The participants in the

control group of stunted

children received weekly

health visits from health

workers and free medical

care.

48 | P a g e


Hamadani et al.,

2006

2 arm Psychosocial stimulation

(improving the mother-child

interaction; and providing

developmentally appropriate

activities for the child) and

nutrition supplementation

through community nutrition

centres

Bangladesh Integrated

Nutrition Program standard

of care, including nutrition

supplementation through

community nutrition centers

Helmizar et al., 2017 4 arm (2X2

factorial)

Received both the parenting

and the nutrition

interventions.

Mothers participated in

weekly parenting classes,

based on a handbook for

psychosocial stimulation

containing 24 age-

appropriate play sessions to

enable mothers to play with

their infants. The main focus

of the play session

programme was to improve

maternal responsivity and

mother-child interaction.

Mothers were expected to

practise the play activities at

home every day. Toys and

picture books were provided

to facilitate this activity at

home.

Participants in the nutrition

component received a

formula food created from

local food sources. Packets

of formula were adjusted for

age group with 200-250 kcal

of energy and 6-8 g of

protein. Caregivers were

provided a handbook with

instructions for preparing

supplements and information

on complementary feeding.


standard of care.

Lozoff et al., 2010 2-arm RCT

(parallel

design)

Home visits were conducted

for infants in the intervention

group by trained professional

educators called 'monitors'.

In the weekly, 1-hour

sessions, the monitors

assessed the family strengths

and challenging areas, set

goals along with the mother,

discussed ECD issues,

offered feedback as mothers

practised activities and

Children in the control group

were administered iron

treatments (30 mg per day)

and had their health, iron

intake and nutrition data

collected in the follow-up

period.

49 | P a g e


helped address any concerns

that the mothers may have

had. All infants received oral

iron treatments (30 mg per

day).

Muhoozi et al., 2017 2 arm Nutritious cooking

demonstrations based on a

nutrition education

curriculum; sanitation and

hygiene activities

(handwashing, use of

toothbrush, etc); and child

stimulation and play.

Standard of care

Nahar et al, 2012

Nahar et al, 2012b

Nahar et al, 2015

5 arms (2x2

factorial with

two control

groups)

Psychosocial stimulation and

food supplementation

Psychosocial stimulation:

play sessions and parental

education using a semi-

structured curriculum

Food supplementation: food

packets were distributed to

children when leaving the

hospital and at each of the

follow-up visits for the first 3

months. The caregivers were

taught about preparation of

the packets.

(1) clinic-control

(2) hospital-control

Powell et al., 2004

Baker-Henningham

et al., 2005

2 arm Stimulation (weekly home

visits by community health

aides; demonstrate play

activities involving the

mother and child; exchange

toys each visit) and nutrition

education

Standard of care (with

nutrition education)

50 | P a g e


Rockers et al., 2016 2 arm (1) Screening and referral for

symptoms of infectious

disease; (2) screening and

referral for acute

malnutrition; (3)

encouragement of the use of

routine health care services

for children; (4) group

meetings addressed different

topics each month, including

parenting skills, child

nutrition and cooking

demonstrations, forms of

play, cognitive stimulation

and language development

activities.

Standard of care

Singla et al., 2015 2 arm 12 60-90 minute-long group

sessions were held that

targeted key messages on

children and maternal well-

being. Topics included use of

play materials, gentle

discipline, consumption of a

diverse diet, hygiene and

sanitation messages. Along

with these discussions, there

were demonstrations for

ways to play and talk to the

child and of food quantities.

Topics for maternal well-

being included strengthening

primary relationships, and

there were sessions for

mothers and fathers,

independently as well as

together.

The waitlist control group

received services from Plan

Uganda as well as nutrition

information around dietary

diversity.

51 | P a g e


Tofail et al, 2013 –

Iron Deficiency

Anaemia (IDA)

2 arm Psychosocial stimulation:

weekly home visits by play

leaders, demonstrations on

how to play with toys and

interact with children;

families receive toys and

books.

Nutrition: one bottle of 35

mL ferrous sulfate syrup was

supplied weekly to the

homes of all children with

IDA for the first 6 months.

Nutrition: one bottle of 35

mL ferrous sulfate syrup was

supplied weekly to the

homes of all children with

IDA for the first 6 months.

Vazir et al., 2013 3 arm The integrated arm of the

trial comprised of a

Responsive Complementary

Feeding and Play Group

programme. Along with the

services that the children

were receiving from

Integrated Child

Development Services,

mothers of children received

11 messages on nutrition

education, 8 on responsive

feeding and 8 on age-

appropriate play-based

stimulation, along with toys,

through 30 home visits.

The nutrition arm of the trial

comprised of a

Complementary Feeding

Group: along with the

services that the children

were receiving from ICDS,

mothers of children in this

arm received 11 messages on

nutrition education through

home visits.

Mothers and infants in the

control group received the

routine Integrated Child

Development Services

services (operating across all

study arms), which includes

centre-based nutritional

supplementation, home-visit

counseling on breastfeeding

and complementary feeding,

monthly growth monitoring,

and non-formal preschool

education for children 3 to 5

years of age.

Waber et al., 1981

Mora et al., 1981

6 arm trial (4

arms

constituted a

2X2 factorial

design: A, D,

A1, D1)

The integrated arm (Arm D1)

received both i) the maternal

education programme and ii)

the weekly food

supplements, beginning in

the 3rd trimester of

pregnancy until the child

reached 3 years of age.

One group (Arm A1) did not

receive supplementation, but

was enrolled in a maternal

education programme with

the goal of increasing

environmental stimulation of

the child group.

Three arms received weekly

food supplementation over

varying increments (Arm B:

from age 6 months to 3

years; Arm C: during 3rd

trimester of pregnancy

through child age 6 months;

Arm D: from beginning of

the 3rd trimester until child

age 3 years). Food

supplements included bread,

52 | P a g e


milk, protein, vitamins, and

minerals.

Yousafzai et al.,

2014

Yousafzai et al.,

2015

4 arms (2x2

factorial)

Responsive Stimulation and

Enhanced Nutrition

Responsive stimulation

intervention: Home visits and

group sessions to promote

caregivers’ sensitivity and

responsiveness by use of

developmentally-appropriate

play activities.

Enhanced nutrition

intervention: responsive

feeding messages,

distribution of a multiple

micronutrient powder for

children aged 6–24 months

to address the prevalent

micronutrient deficiencies in

the population,

nutrition/heath messages

Lady Health Worker

Programme - standard of

care

The meta-analyses for child outcomes are organized in three GRADE tables: (1)

combined nutrition and caregiving interventions versus standard of care; (2) combined nutrition

and caregiving versus caregiving interventions alone; and (3) combined nutrition and caregiving

interventions versus nutrition interventions alone. The child outcomes are cognitive

development, language development, motor development, socioemotional development, HAZ,

WAZ, weight-for-height (WHZ), and any other child nutrition or health outcome.


For child cognitive, language and motor development outcomes the majority of the evaluations

employed BSID. Three studies from Jamaica employed the Griffiths Mental Development Scale

(Powell et al., 2004; Gardner et al., 2005; Grantham-McGregor et al., 1991). Two studies used

caregiver reports (Muhoozi et al., 2017; Rockers et al., 2016). Only two programmes evaluated

impact on socioemotional development, and both used the BSID.

Cognitive development:

• For combined nutrition and caregiving interventions versus standard of care, the pooled

results showed a significant improvement in cognitive development (SMD = 0.57, 95% CI:

0.32 to 0.88, n = 14). The overall quality of evidence was graded as low.

• For combined nutrition and caregiving versus caregiving interventions alone, the pooled

results showed no significant improvement in cognitive development (SMD = 0.10, 95% CI:

-0.12 to 0.32, n = 6). The overall quality of evidence was graded as low.

• For combined nutrition and caregiving versus nutrition interventions alone, the pooled results

showed a significant improvement in cognitive development (SMD = 0.45, 95% CI: 0.22 to

0.67, n = 9). The overall quality of evidence was graded as low.

Language development:


results showed a significant improvement in language development (SMD = 0.40, 95% CI:

0.17, 0.63, n = 10). The overall quality of evidence was graded as low.


results showed no significant improvement in language development (SMD = 0.01, 95% CI:

-0.09 to 0.10, n = 6). The overall quality of evidence was graded as moderate.

• For combined nutrition and caregiving intervention versus nutrition interventions alone, the

pooled results showed a significant improvement in language development (SMD = 0.21,

95% CI: 0.13 to 0.28, n = 6). The overall quality of evidence was graded as moderate.

54 | P a g e

Motor development:


results showed a significant improvement in motor development (SMD = 0.4, 95% CI: 0.26

to 0.53, n = 10). The overall quality of evidence was graded as low.


results showed no significant improvement in motor development (SMD = 0.18, 95% CI: -

0.06 to 0.42, n = 6). The overall quality of evidence was graded as low.

• For combined nutrition and caregiving interventions versus nutrition interventions alone, the

pooled results showed no significant improvement in motor development (SMD = 0.14, 95%

CI: 0.07 to 0.22, n = 9). The overall quality of evidence was graded as high.

Socioemotional development:


results showed no significant improvement in socioemotional development (SMD = 0.09,

95% CI: -0.11, 0.30, n = 2). The overall quality of evidence was graded as low.

• For combined nutrition and caregiving versus caregiving interventions alone, Yousafzai and

colleagues (2014) had an effect size of 0.11, 95% CI: -0.04 to 0.26. The overall quality of


• For combined nutrition and caregiving interventions versus nutrition interventions alone,

Yousafzai and colleagues (2014) had an effect size of -0.09, 95% CI: 0.24 to 0.07. The


Overall, no significant benefits were found on child growth outcomes.

HAZ:


results showed SMD = -0.13, 95% CI: -0.31 to 0.05, n = 9. The overall quality of evidence

was graded as low.


results showed SMD = -0.21, 95% CI: -0.60 to 0.19, n = 4. The overall quality of evidence

was graded as low.


pooled results showed SMD = -0.42, 95% CI: -0.85 to 0.01, n = 4. The overall quality of


WAZ:


results showed SMD = 0.06, 95% CI: -0.02 to 0.13, n = 7. The overall quality of evidence

was graded as high.

55 | P a g e


results showed SMD = 0.07, 95% CI: -0.04 to 0.17, n = 3). The overall quality of evidence

was graded as moderate.


pooled results showed SMD = 0.06, 95% CI: -0.02 to 0.14, n = 4. The overall quality of

evidence was graded as moderate.

WHZ:


results showed SMD = 0.20, 95% CI: 0.05 to 0.34, n = 6. The overall quality of evidence was

graded as moderate.


results showed SMD = 0.16, 95% CI: 0.03 to 0.29, n = 4. The overall quality of evidence was

graded as moderate.


pooled results showed SMD = 0.17, 95% CI: -0.04 to 0.38, n = 5. The overall quality of


Impact on other child nutrition and child health outcomes: In general, child nutrition and health

outcomes were not commonly or consistently measured across studies; therefore, these outcomes

were not included in the meta-analysis. Aboud and Akhter (2011) reported significant

improvements in the intervention groups (receiving combined nutrition and caregiving

interventions) on mouthfuls eaten and handwashing practices. Vazir and colleagues (2013) found

children in the nutrition arms (with or without early learning interventions) of the trial

significantly improved dietary intake as a result of nutrition education intervention exposure.

Yousafzai and colleagues (2015) reported children exposed to nutrition interventions (with or

without responsive caregiving and early learning) had significantly improved age-appropriate

breast feeding practices, and children exposed to the caregiving interventions (with or without

the nutrition interventions) had significantly improved minimal acceptable diet and meal

frequency . Three studies found no effects on reducing sickness (Aboud et al., 2013; Menon et

al., 2016; Singla et al., 2015), while another study found reductions in diarrhoea and acute

respiratory illness (with or without responsive caregiving and early learning) (Yousafzai et al.,

2014).

Subgroup analyses

We analysed the interventions by whether they targeted malnourished children compared

with universal implementation (no targeting). The findings are shown in Tables 10-12. In each

56 | P a g e

comparison, we find the effect size for development outcomes are higher for malnourished

children than the universally-implemented studies, with mixed results on growth outcomes.

Table 10. Combined responsive caregiving and early learning versus standard of care

Outcome Overall Targeted: malnourished Universal

SMD 95% CI N SMD 95% CI N SMD 95% CI N

Cognitive

development

0.57 0.32, 0.82 13 0.63 0.34, 0.92 6 0.52 0.15, 0.88 7

Language

development

0.40 0.17, 0.63 10 0.56 0.32, 0.81 3 0.35 0.07, 0.63 7

Motor development 0.40 0.26, 0.53 10 0.38 0.13, 0.64 5 0.41 0.25, 0.58 5

Attachment

Socioemotional

development

0.09 -0.11, 0.30 1

HAZ -0.13 -0.31, 0.05 9 -0.36 -0.88, 0.15 2 -0.07 -0.25, 0.11 7

WAZ 0.06 -0.02, 0.13 7 0.00 -0.14, 0.14 2 0.08 -0.01, 0.17 5

WHZ 0.20 0.05, 0.34 6 0.11 -0.10, 0.31 3 0.25 0.04, 0.46 3

Table 11. Combined responsive caregiving and early learning versus caregiving alone



Cognitive

development

0.10 -0.12, 0.32 6 0.32 -0.03, 0.66 4 -0.13 -0.24, -0.02 2

Language

development

0.01 -0.09, 0.10 6 0.26 -0.12, 0.63 2 -0.01 -0.11, 0.09 4

Motor

development

0.18 -0.06, 0.42 6 0.42 -0.26, 1.09 3 0.06 -0.14, 0.25 3

Attachment

Socioemotional

development

0.11 -0.04, 0.26 1

HAZ -0.21 -0.60, 0.19 4 -0.83 -1.19, -0.46 1 0.00 -0.25, 0.25 3

WAZ 0.07 -0.04, 0.17 3 0.00 -0.15, 0.15 1 0.12 -0.02, 0.27 2

WHZ 0.16 0.03, 0.29 4 0.09 -0.19, 0.38 2 0.18 0.03, 0.32 2

Table 12. Combined responsive caregiving and early learning versus nutrition alone



Cognitive

development

0.45 0.22, 0.67 9 0.61 0.18, 1.05 6 0.23 0.13, 0.33 3

Language

development

0.21 0.13, 0.28 6 0.43 -0.20, 1.07 2 0.21 0.15, 0.27 4

Motor

development

0.14 0.07, 0.22 9 0.07 -0.16, 0.30 4 0.17 0.11, 0.23 6

57 | P a g e



Attachment

Socioemotional

development

-0.08 -0.24, 0.07 1

HAZ -0.42 -0.85, 0.01 4 -1.28 -1.65, -0.91 1 -0.14 -0.41, 0.14 3

WAZ 0.06 -0.02, 0.14 4 0.15 -0.00, 0.31 1 0.04 -0.03, 0.10 3

WHZ 0.17 -0.04, 0.38 5 0.13 -0.14, 0.40 2 0.20 -0.09, 0.48 3


While there is no consistent evidence for additive benefits on single outcomes, combined

caregiving and nutrition can impact a number of child and caregiving outcomes. There does not

appear to be any significant evidence for adverse effects. Cost of interventions is not reported,

but potential cost savings for programmes may be possible when using the same platform and

delivery agent to deliver integrated nurturing care for children.

In conclusion, the evidence from LMICs suggests that combined caregiving and nutrition

interventions are significantly effective on child cognitive, language and motor development

compared with usual care, and on child cognitive and language development compared with

nutrition alone. No benefits are observed on growth outcomes. Among malnourished

populations, combined caregiving and nutrition interventions are significantly effective on child

cognitive, language and motor development compared with usual care, and on child cognitive

development compared with nutrition alone. More research is needed on how to optimize the

combined nutrition and caregiving strategies.

Research gaps

In undertaking the systematic reviews, the following research gaps were identified by the

research team.

1. The largest group of studies were for (additional analysis) 2.1 (combined responsive

caregiving and promotion of early learning) with good global representation. However, PICO

1 (responsive caregiving alone) had limited intervention research in LMICs. Similarly, PICO

4 (integrated caregiving and nutrition programmes) was largely focused on undernutrition in

LMICs and did not address the growing challenges in many countries on child overnutrition.

2. Few studies report on caregiving-related outcomes, which are critical for understanding

processes for the effectiveness of caregiving/parenting programmes on child outcomes

(Jeong et al., 2018).

3. Few studies reported adequate information on programme characteristics to enable analysis

on actual implementation processes, rather than intended implementation processes (e.g.

dosage, behaviour change techniques). These data are essential to inform evidence-based

implementation planning of nurturing care programmes.

4. There is little information on caregiving and overnutrition.

5. Few studies reported findings on subgroups to determine whether interventions were more or

less effective for particular groups within the population (e.g. child and caregiver

characteristics).

6. Limited data on cost were reported in relation to the interventions. Policy makers require cost

information to plan programmes, and more research is required on costing of interventions.

7. Many tools employed to assess both child and caregiving outcomes are unstandardized,

making it difficult to assess a specific construct of development. Reporting about the

reliability and validity of adapted tools is limited.

8. Definitions for interventions are variable, making comparisons challenging. The systematic

review team defined common characteristics for interventions categorized as responsive

caregiving, early learning promotion, and support for socioemotional and behavioural

development. Clear intervention reporting guidelines would be helpful for the multi-

disciplinary research community working in the field of ECD.

9. Data from large scale studies are limited.

10. Data that report on caregivers, other than mothers, and measure outcomes on other caregivers

are limited.

59 | P a g e

References

Cited in introduction and methods

Aboud FE, Yousafzai AK. Very early childhood development. Reproductive, Maternal,

Newborn, and Child Health. 2016: 241.

Aboud FE, Yousafzai AK. Global health and development in early childhood. Ann Rev Psychol.

2015; 66:433-57.

Black MM, Walker SP, Fernald LC, et al. Early childhood development coming of age: science

through the life course. Lancet 2017; 389(10064): 77-90.

Britto PR, Lye SJ, Proulx K, et al. Nurturing care: promoting early childhood development.

Lancet. 2017; 389(10064): 91-102.

Britto P, Ponguta L, Reyes C, Karnati R. A systematic review of parenting programmes for

young children in low-and middle-income countries. New York, NY: United Nations Children’s

Fund; 2015.

Eshel N, Daelmans B, Mello MCd, Martines J. Responsive parenting: interventions and

outcomes. Bull World Health Org. 2006; 84(12): 991-8.

Fergusson DM, Grant H, Horwood LJ, Ridder EM. Randomized trial of the Early Start program

of home visitation. Pediatrics. 2005; 116(6): e803-9.

Jeong J, Pitchik HO, Yousafzai AK. Stimulation interventions and parenting in low- and middle-

income countries: a meta-analysis. Pediatrics. 2018; e20173510.

Lu C, Black MM, Richter LM. Risk of poor development in young children in low-income and

middle-income countries: an estimation and analysis at the global, regional, and country level.

Lancet Global Health. 2016; 4(12): e916-e22.

Nelson CA. The neurobiological bases of early intervention. Handbook of early childhood

intervention. 2000; 2: 204-27.

Mol SE, Bus AG, De Jong MT, Smeets DJ. Added value of dialogic parent–child book readings:

a meta-analysis. Early Education and Development. 2008; 19(1): 7-26.

Richter LM, Daelmans B, Lombardi J, et al. Investing in the foundation of sustainable

development: pathways to scale up for early childhood development. Lancet. 2017; 389(10064):

103-18.

60 | P a g e

Tanner‐Smith EE & Tipton E. Robust variance estimation with dependent effect sizes: practical

considerations including a software tutorial in Stata and SPSS. Research Synthesis

Methods. 2014; 5(1): 13-30.

Studies/records included in review

Aboud FE, Akhter S. A cluster-randomized evaluation of a responsive stimulation and feeding

intervention in Bangladesh. Pediatrics. 2011; 127(5): e1191-7.

Aboud FE, Singla DR, Nahil MI, Borisova I. Effectiveness of a parenting program in Bangladesh

to address early childhood health, growth and development. Soc Sci Med. 2013; 97: 250-8.

Attanasio OP, Fernández C, Fitzsimons EO, Grantham-McGregor SM, Meghir C, Rubio-Codina

M. Using the infrastructure of a conditional cash transfer program to deliver a scalable integrated

early child development program in Colombia: cluster randomized controlled trial. BMJ. 2014;

349: g5785.

Baker-Henningham H, Powell C, Walker S, Grantham-McGregor S. The effect of early

stimulation on maternal depression: a cluster randomised controlled trial. Arch Dis Child. 2005;

90(12): 1230-4.

Barlow J., et al. Role of home visiting in improving parenting and health in families at risk of

abuse and neglect: results of a multicentre randomised controlled trial and economic evaluation.

Arch Dis Child. 2007; 92(3): 229-33.

Barlow A, Mullany B, Neault N, et al. Paraprofessional-delivered home-visiting intervention for

American Indian teen mothers and children: 3-year outcomes from a randomized controlled trial.

Am J Psychiatry. 2015; 172(2): 154-62.

Barrera M. E., et al. Early home intervention with low-birth-weight infants and their parents.

Child Dev. 1986; 57(1): 20-33.

Breitenstein SM, Gross D, Fogg L, Ridge A, Garvey C, Julion W et al. The Chicago Parent

Program: comparing 1-year outcomes for African American and Latino parents of young

children. Res Nurs Health. 2012; 35(5): 475-89.

Brooks-Gunn J, et al. Effects of early intervention on cognitive function of low birth weight

preterm infants. Pediatrics. 1992; 120: 350-9.

Caldera D, Burrell L, Rodriguez K, Crowne SS, Rohde C, Duggan A. Impact of a statewide

home visiting program on parenting and on child health and development. Child Abuse Negl.

2007; 31(8): 829-52.

61 | P a g e

Caughy MOB, et al. The effects of the Healthy Steps for Young Children Program: results from

observations of parenting and child development." Early Childhood Research Quarterly. 19(4):

611-30.

Chang SM, Grantham-McGregor SM, Powell CA, et al. Integrating a parenting intervention with

routine primary health care: a cluster randomized trial. Pediatrics. 2015; 136(2): 272-80.

Cheng S, et al. The effectiveness of early intervention and the factors related to child behavioural

problems at age 2: a randomized controlled trial. Early Hum Dev. 2007; 83(10): 683-91.

Constantino JN, et al. Supplementation of urban home visitation with a series of group meetings

for parents and infants: results of a "real-world" randomized, controlled trial. Child Abuse Negl.

25(12): 1571-81.

Cooper PJ, Tomlinson M, Swartz L, et al. Improving quality of mother-infant relationship and

infant attachment in socioeconomically deprived community in South Africa: randomised

controlled trial. BMJ. 2009; 338(7701): 997.

Cronan T A, et al. The effects of a community-based literacy program on young children's

language and conceptual development. Am J Community Psychol. 1996; 24(2): 51-272.

Dishion TJ, Shaw D, Connell A, Gardner F, Weaver C, Wilson M. The family check-up with

high-risk indigent families: preventing problem behavior by increasing parents' positive behavior

support in early childhood. Child Dev. 2008; 79(5): 1395-414.

Dozier M, et al. Effects of a foster parent training program on young children's attachment

behaviors: preliminary evidence from a Randomized Clinical Trial. Child Adolesc Social Work

J. 26(4): 321-32.

Drotar D, Robinson J, Jeavons L, Lester Kirchner H. A randomized, controlled evaluation of

early intervention: the Born to Learn curriculum. Child Care Health Dev. 2009; 35(5): 643-9.

Fergusson DM, Grant H, Horwood J, Ridder EM Randomized trial of the Early Start Program of

Home Visitation. Pediatricts. 2005; 116(6): e803-e809.

Fernald LC, Kagawa R, Knauer HA, Schnaas L, Guerra AG, Neufeld LM. Promoting child

development through group-based parent support within a cash transfer program: experimental

effects on children’s outcomes. Developmental psychology. 2017; 53(2): 222.

Field T, et al. Effects of parent training on teenage mother and their infants. Pediatrics. 69(6):

703-7.

Frongillo EA, Nguyen PH, Saha KK, et al. Large-scale behavior-change initiative for infant and

young child feeding advanced language and motor development in a cluster-randomized program

evaluation in Bangladesh. J Nutr. 2017; 147(2): 256-63.

62 | P a g e

Gardner JM, Powell CA, Baker-Henningham H, Walker SP, Cole TJ, Grantham-McGregor SM.

Zinc supplementation and psychosocial stimulation: effects on the development of

undernourished Jamaican children. Am J Clin Nutr. 2005; 82(2): 399-405.

Goldfeld S, Napiza N, Quach J, Reilly S, Ukoumunne OC, Wake M. Outcomes of a universal

shared reading intervention by 2 years of age: the Let's Read trial. Pediatrics. 2011; 127(3): 445-

53.

Goodson BD, Layzer JI, Pierre RG, Bernstein LS, Lopez M. Effectiveness of a comprehensive,

five-year family support program for low-income children and their families: findings from the

Comprehensive Child Development Program. Early Childhood Research Quarterly. 2000; 15(1):

5-39.

Gowani S, Yousafzai AK, Armstrong R, Bhutta ZA. Cost effectiveness of responsive

stimulation and nutrition interventions on early child development outcomes in Pakistan. Ann N

Y Acad Sci. 2014; 1308: 149-161.

Grantham-McGregor SM, Powell CA, Walker SP, Himes JH. Nutritional supplementation,

psychosocial stimulation, and mental development of stunted children: the Jamaican Study.

Lancet. 1991; 338(8758): 1-5.

Gross D, Garvey C, Julion W, Fogg L, Tucker S, Mokros H. Efficacy of the Chicago parent

program with low-income African American and Latino parents of young children. Prevention

science: the official journal of the Society for Prevention Research. 2009; 10(1): 54-65.

Guedeney A, et al. Impact of a randomized home-visiting trial on infant social withdrawal in the

CAPEDP prevention study. Infant Mental Health Journal. 2013; 34(6): 594-601.

Guttentag CL, Landry SH, Williams JM, et al. "My Baby & Me": effects of an early,

comprehensive parenting intervention on at-risk mothers and their children. Dev Psychol. 2014;

50(5): 1482-96.

Hamadani JD, Huda SN, Khatun F, Grantham-McGregor SM. Psychosocial stimulation

improves the development of undernourished children in rural Bangladesh. J Nutr. 2006;

136(10): 2645-52.

Hartinger SM, Lanata CF, Hattendorf J, et al. Impact of a child stimulation intervention on early

child development in rural Peru: a cluster randomised trial using a reciprocal control design. J

Epidemiol Community Health. 2017; 71(3): 217-24.

Heinicke CM, et al. Relationship-based intervention with at-risk mothers: Outcome in the first

year of life. Infant Mental Health Journal. 1999; 20(4): 349-74.

Helmizar H, Jalal F, Lipoeto NI, Achadi EL. Local food supplementation and psychosocial

stimulation improve linear growth and cognitive development among Indonesian infants aged 6

to 9 months. Asia Pacific Journal of Clinical Nutrition. 2017; 26(1): 97.

63 | P a g e

Heubner CE Promoting toddlers' language development through community-based intervention.

Journal of Applied Developmental Psychology. 21(5): 513-35.

Hiscock H, Bayer JK, Price A, Ukoumunne OC, Rogers S, Wake M. Universal parenting

programme to prevent early childhood behavioural problems: cluster randomised trial. BMJ

2008; 336(7639): 318-21.

Hiscock H, Gulenc A, Ukoumunne OC, et al. Preventing preschool mental health problems:

population-based cluster randomized controlled trial. J Dev Behav Pediatr. 2018; 39(1): 55-65.

Hutchings J, et al. Evaluating the Incredible Years Toddler Parenting Programme with parents of

toddlers in disadvantaged (Flying Start) areas of Wales. Child Care Health Dev. 2016; 43(1):

104-13.

Jacobs F, Easterbrooks MA, Goldberg J, et al. Improving adolescent parenting: results from a

randomized controlled trial of a home visiting program for young families. Am J Public Health.

2016; 106(2): 342-9.

Jin X, et al. "Care for Development" intervention in rural China: a prospective follow-up study. J

Dev Behav Pediatr. 28(3): 213-8.

Johnson DL, et al. TheHouston Parent-childDevelopment Center: a parent education program for

Mexican-American families. Amer J Orthopsychiat. 44(1): 121-8.

Juffer F, et al. Early intervention in adoptive families: supporting maternal sensitive

responsiveness, infant-mother attachment, and infant competence. J Child Psychol Psychiatry.

38(8): 1039-50.

Kaaresen PI, et al. A randomized controlled trial of an early intervention program in low birth

weight children: outcome at 2 years. Early Hum Dev. 2008; 84(3): 201-9.

Kalinauskiene L, et al. Supporting insensitive mothers: the Vilnius randomized control trial of

video-feedback intervention to promote maternal sensitivity and infant attachment security.

Child Care Health Dev.2009; 35(5): 613-23.

Kaminski JW, Perou R, Visser SN, et al. Behavioral and socioemotional outcomes through age 5

years of the legacy for children public health approach to improving developmental outcomes

among children born into poverty. Am J Public Health. 2013; 103(6): 1058-66.

Kitzman H, Olds DL, Henderson CR Jr, et al. Effect of prenatal and infancy home visitation by

nurses on pregnancy outcomes, childhood injuries, and repeated childbearing. A randomized

controlled trial. JAMA . 1997; 278(8): 644-52.

Kochanska G, et al. Promoting toddlers' positive social-emotional outcomes in low-income

families: a play-based experimental study. J Clin Child Adolesc Psychol. 2013; 42(5): 700-12.

64 | P a g e

Kyno NM, et al. Effect of an early intervention programme on development of moderate and

late preterm infants at 36 months: a randomized controlled study. Infant Behav Dev. 2012; 35(4):

916-26.

Leung C, Tsang S, Li B. Efficacy of Fun to Learn for the Young Program: randomized

controlled trial. J Child Fam Stud. 2017a; 26: 2865-78.

Leung C, et al. Evaluation of Parent and Child Enhancement (PACE) program: randomized

controlled trial. Research on Social Work Practice. 2017b; 27(1): 19-35.

Love JM, Kisker EE, Ross C, et al. The effectiveness of early head start for 3-year-old children

and their parents: lessons for policy and programs. Dev Psychol. 2005; 41(6): 885-901.

Lozoff B, Smith JB, Clark KM, Perales CG, Rivera F, Catillo M. Home intervention improves

cognitive and social-emotional scores in iron-deficient anemic infants. Pediatrics. 2010; 126(4):

e884-e894.

Madden J, et al. Home again: effects of the Mother-Child Home Program on mother and child.

Child Dev. 1984; 55(2): 636-47.

McGillion M, et al. A randomised controlled trial to test the effect of promoting caregiver

contingent talk on language development in infants from diverse socioeconomic status

backgrounds. J Child Psychol Psychiatry. 2017; 58(10): 1122-31.

Mendelsohn AL, Valdez P, Flynn V, Foley G, Berkule S, Tomopoulos S, et al. Use of

videotaped interactions during pediatric well-child care: impact at 33 months on parenting and

on child development. Journal of Developmental and Behavioral Pediatrics. 2007; 28:206–12.

Menon P, Nguyen PH, Saha KK, Khaled A, Sanghvi T, Baker J, et al. Combining intensive

counseling by frontline workers with a nationwide mass media campaign has large differential

impacts on complementary feeding practices but not on child growth: results of a cluster-

randomized program evaluation in Bangladesh–3. Journal of Nutrition. 2016; 146(10): 2075-84.

Mora J, Herrera M, Suescun J, De Navarro L, Wagner M. The effects of nutritional

supplementation on physical growth of children at risk of malnutrition. Am J Clinical Nutrition.

1981; 34(9): 1885-92.

Muhoozi GKM, Atukunda P, Diep LM, et al. Nutrition, hygiene, and stimulation education to

improve growth, cognitive, language, and motor development among infants in Uganda: a

cluster-randomized trial. Matern Child Nutr. 2017;

Murray L, De Pascalis L, Tomlinson M, et al. Randomized controlled trial of a book-sharing

intervention in a deprived South African community: effects on carer-infant interactions, and

their relation to infant cognitive and socioemotional outcome. J Child Psychol Psychiatry. 2016;

57(12): 1370-9.

65 | P a g e

Nahar B, Hossain MI, Hamadani JD, et al. Effects of a community-based approach of food and

psychosocial stimulation on growth and development of severely malnourished children in

Bangladesh: a randomised trial. Eur J Clin Nutr. 2012a; 66(6): 701-9.

Nahar B, Hossain MI, Hamadani JD, Ahmed T, Grantham-McGregor S, Persson LA. Effects of

psychosocial stimulation on improving home environment and child-rearing practices: results

from a community-based trial among severely malnourished children in Bangladesh. BMC

Public Health. 2012b; 12: 622.

Nahar B, Hossain I, Hamadani JD, Ahmed T, Grantham‐ McGregor S, Persson LA. Effect of a

food supplementation and psychosocial stimulation trial for severely malnourished children on

the level of maternal depressive symptoms in Bangladesh. Child: care, health and development.

2015; 41(3): 483-93.

Nair MKC, Philip E, Jeyaseelan L, George B, Mathews S, Padma K. Effect of Child

Development Centre Model early stimulation among at-risk babies-a randomized controlled

trial. Indian Pediatrics. 2009; 46: S20-S26.

Norr KF, Crittenden KS, Lehrer EL, et al. Maternal and infant outcomes at one year for a nurse-

health advocate home visiting program serving African Americans and Mexican Americans.

Public Health Nurs. 2003; 20(3): 190-203.

Olds DL, et al. Preventing child abuse and neglect: a randomized trial of nurse home visitation.

Pediatrics. 1986. 78(1): 65-78.

Olds DL, Robinson J, O'Brien R, et al. Home visiting by paraprofessionals and by nurses: a

randomized, controlled trial. Pediatrics. 2002; 110(3): 486-96.

Pontoppidan M, et al. The Incredible Years Parents and Babies Program: a pilot randomized

controlled trial." PLoS One. 2016;11(12): e0167592.

Powell C, Grantham McGregor S. Home visiting of varying frequency and child development.

Pediatrics. 1989; 84(1): 157-64.

Powell C, Baker-Henningham H, Walker S, Gernay J, Grantham-McGregor S. Feasibility of

integrating early stimulation into primary care for undernourished Jamaican children: cluster

randomised controlled trial. BMJ. 2004; 329(7457): 89.

Rauh VA, et al. Minimizing adverse effects of low birthweight: four-year results of an early

intervention program. Child Dev. 1988; 59(3): 544-53.

Robling M, Bekkers MJ, Bell K, et al. Effectiveness of a nurse-led intensive home-visitation

programme for first-time teenage mothers (Building Blocks): a pragmatic randomised controlled

trial. Lancet. 2016; 387(10014): 146-55.

66 | P a g e

Rockers PC, Fink G, Zanolini A, et al. Impact of a community-based package of interventions on

child development in Zambia: a cluster-randomised controlled trial. BMJ Global Health. 2016;

1(3): e000104.

Roggman LA, et al. Keeping kids on track: mpacts of a parenting-focused Early Head Start

program on attachment security and cognitive development. Early Education and Development.

2009; 20(6): 920-41.

Santelices MP, et al. Promoting secure attachment: Evaluation of the effectiveness of an early

intervention pilot programme with mother–infant dyads in Santiago, Chile. Child: Care, Health

and Development. 2011; 37(2): 203-10.

Sawyer MG, Reece CE, Bowering K, et al. Nurse-moderated internet-based support for new

mothers: non-inferiority, randomized controlled trial. J Med Internet Res. 2017; 19(7): e258.

Scarr S, McCartney K. Far from home: an experimental evaluation of the Mother-Child Home

Program in Bermuda. Child Development. 1988; 59(3): 531-43.

Schwarz DF, O'Sullivan AL, Guinn J, et al. Promoting early intervention referral through a

randomized controlled home-visiting program. Journal of Early Intervention. 2012; 34(1): 20-39.

Shaw DS, Connell A, Dishion TJ, Wilson MN, Gardner F. Improvements in maternal depression

as a mediator of intervention effects on early childhood problem behavior. Dev Psychopathol.

2009; 21(2): 417-39.

Singla DR, Kumbakumba E, Aboud FE. Effects of a parenting intervention to address maternal

psychological wellbeing and child development and growth in rural Uganda: a community-

based, cluster randomised trial. Lancet Global Health. 2015; 3(8): e458-e69.

Spieker SJ, Oxford ML, Kelly JF, Nelson EM, Fleming CB. Promoting first relationships:

randomized trial of a relationship-based intervention for toddlers in child welfare. Child

Maltreatment. 2012; 17(4): 271-86.

Tofail F, Hamadani JD, Mehrin F, Ridout DA, Huda SN, Grantham-McGregor SM. Psychosocial

stimulation benefits development in nonanemic children but not in anemic, iron-deficient

children–3. Journal of Nutrition. 2013; 143(6): 885-93.

Vally Z, et al. The impact of dialogic book‐sharing training on infant language and attention: a

randomized controlled trial in a deprived South African community. Journal of Child Psychology

and Psychiatry. 2015;56(8): 865-73.

Van Zeijl J, Mesman J, Van IJzendoorn MH, et al. Attachment-based intervention for enhancing

sensitive discipline in mothers of 1-to 3-year-old children at risk for externalizing behavior

problems: a randomized controlled trial. Journal of Consulting and Clinical Psychology. 2006;

74(6): 994.

67 | P a g e

Vazir S, Engle P, Balakrishna N, et al. Cluster-randomized trial on complementary and

responsive feeding education to caregivers found improved dietary intake, growth and

development among rural Indian toddlers. Matern Child Nutr. 2013; 9(1): 99-117.

Velderman MK, et al. Effects of attachment-based interventions on maternal sensitivity and

infant attachment: differential susceptibility of highly reactive infants. J Fam Psychol. 2006;

20(2): 266-74.

Waber DP, Vuori-Christiansen L, Ortiz N, et al. Nutritional supplementation, maternal

education, and cognitive development of infants at risk of malnutrition. AmJ Clin Nut. 1981;

34(4): 807-13.

Wagner M, et al. The effectiveness of the Parents as Teachers program with low-income parents

and children. Topics in Early Childhood Special Education. 2002; 22(2): 67-81.

Wake M, Tobin S, Girolametto L, et al. Outcomes of population based language promotion for

slow to talk toddlers at ages 2 and 3 years: Let’s Learn Language cluster randomised controlled

trial. BMJ. 2011; 343(7821): 1-10.

Walker SP, Powell CA, Grantham-McGregor SM, Himes JH, Chang SM. Nutritional

supplementation, psychosocial stimulation, and growth of stunted children: the Jamaican study.

Am J Clin Nut. 1991; 54(4): 642-8.

Walker SP, Chang SM, Powell CA, Grantham-McGregor, SM. Psychosocial intervention

improves the development of term low-birth-weight infants. Community and International

Nutrition. 2004; 134(6): 1417-23.

Walkup JT, et al. Randomized controlled trial of a paraprofessional-delivered in-home

intervention for young reservation-based American Indian mothers." J Am Acad Child Adolesc

Psychiatry. 2009;48(6): 591-601.

Wallander JL, Bann CM, Biasini FJ, et al. Development of children at risk for adverse outcomes

participating in early intervention in developing countries: a randomized controlled trial. J Child

Psychol Psychiatry. 2014; 55(11): 1251-9.

Wasik BH et al. A longitudinal study of two early intervention strategies: Project CARE. Child

Dev. 1990; 61(6): 1682-96.

Weisleder A, Cates CB, Dreyer BP, et al. Promotion of positive parenting and prevention of

socioemotional disparities. Pediatrics. 2016; 137(2): e20153239.

Whitt JK, Casey PH. The mother–infant relationship and infant development: the effect of

pediatric intervention. Child Development. 1982; 53(4): 948-56.

Yousafzai AK, Rasheed MA, Rizvi A, Armstrong R, Bhutta ZA. Effect of integrated responsive

stimulation and nutrition interventions in the Lady Health Worker programme in Pakistan on

68 | P a g e

child development, growth, and health outcomes: a cluster-randomised factorial effectiveness

trial. Lancet. 2014; 384(9950): 1282-93.

Yousafzai AK, Rasheed MA, Rizvi A, Armstrong R, Bhutta ZA. Parenting skills and emotional

availability: an RCT. Pediatrics. 2015; 135(5): e1247-57.

69 | P a g e

Appendix A: Glossary

• Attachment: an emotional bond between an infant and one or more adults. The infant will

approach these individuals in times of distress, particularly during the phase of infant

development when the presence of strangers induces anxiety. In addition, the infant is

distressed if separated from attachment figures.1

o Attachment status: a description of an infant’s attachment as being either secure or

insecure.

▪ Secure attachment: a child who is securely attached actively explores the

environment in the presence of the caregiver, is visibly upset by separation,

and greets the mother warmly when they are reunited.

▪ Insecure attachment: attachment that takes one of three forms: avoidant

attachment, anxious-resistant attachment and disorganized/disoriented

attachment.

• Attunement: an empathic responsiveness between two individuals, described by Daniel

Stern as the ‘performance of behaviours that express the quality of feeling of a shared affect

state.’2

• Behaviour problems:

o Externalizing: “behaviour problems that are manifested in children’s outward

behaviour and reflect the child negatively acting on the external environment. Other

terms to describe externalizing behaviour problems include ‘conduct problems,’

‘antisocial,’ and ‘under-controlled.’3

o Internalizing: “behaviour problems such as withdrawn, anxious, inhibited, and

depressed behaviours that more centrally affect the child’s internal psychological

environment rather than the external world.”3

• Bundling: combining two or more services in a single programme, with the goal of

maintaining or enhancing the benefits of existing services and gaining additional benefits

from the new intervention.4

• Developmental potential: ability to think, learn, remember, relate, and articulate ideas

appropriate to age and level of maturity; an estimated 39% of the world’s children under age

five years do not attain this potential.5

1 Richter L. The importance of caregiver-child interactions for the survival and health development of young children: a review. Geneva: World Health Organization; 2004. 2 Stern DN. The interpersonal world of the infant: a view from psychoanalysis and developmental psychology. New York: Basic Books; 1985, p.

142; cited in Richter L. 2004. 3 Liu J. Childhood externalizing behavior: theory and implications. Journal of Child and Adolescent Psychiatric Nursing. 2004;17(3): 93–103. 4 Alderman H. Early childhood development: does bundling services for young children and their families reduce costs? Brookings: Education

Plus Development; 2015. 5 Grantham-McGregor S. et al. Developmental potential in the first 5 years for children in developing countries. Lancet. 2007; 369:9555:60-70.

70 | P a g e

• Dialogic book-sharing (dialogic reading): stimulation package designed according to the

following three principles: (a) use of techniques by the parent to encourage the child to talk

about pictured materials; (b) informative feedback by incorporating expansions, corrective

modelling, and other forms that highlight differences between what the child has said and

what he/she might have said; and (c) an adaptive parent sensitive to the child’s developing

abilities.1

• Depression: an affective disorder characterized by a sense of inadequacy, feelings of

despondency or hopelessness, a decrease in activity and/or reactivity, pessimism, sadness,

irritability, changes in appetite and sleep patterns, and poor concentration.

• Early childhood development (ECD): refers to the physical, socioemotional, cognitive, and

motor development between 0-8 years of age.

• Emotional availability: refers to the ability of the caregiver and child to share a healthy

emotional connection and the quality of emotional exchanges between caregivers and

children. Encompasses both emotional signaling and emotional understanding, as well as the

emotional accessibility of one to the other. The emotionally available dyad is one in which

both mother and infant recognize the other partner’s signals and affirm them.2

• Home visiting programmes: involve visits by nurses to parents and children in their homes

to prevent child maltreatment and promote positive infant, child and parental development by

providing support, education and information.3

• Integration: same as bundling; combining two or more services in a single intervention,

with the goal of maintaining or enhancing the benefits of existing services and gaining some

benefit from the new programme.

• Interventions: attempts to influence or change the course of events by providing care or

information or otherwise manipulating a situation.

• Macronutrients: include carbohydrates, proteins, and fats. Consumed in relatively large

quantities and are important to child linear growth and mental development.4

• Malnutrition: results from deficiencies, excesses or imbalances in the consumption of

macro- and/or micronutrients. Malnutrition may be an outcome of food insecurity, or it may

relate to non-food factors, such as inadequate care practices for children, insufficient health

services, and/or an unhealthy environment.5

o Pediatric malnutrition (undernutrition): an imbalance between nutrient

requirements and intake that results in cumulative deficits of energy, protein or

1 Mol SE, et al. Added value of dialogic parent-child book readings: a meta-Analysis. Early Education and Development. 2008;19:1, 7-26. 2 Saunders H, Kraus A, Barone L, Biringen Z. Emotional availability: theory, research, and intervention. Front Psychol. 2015;6:1069; Bornstein

MH et al. Emotional relationships between mothers and infants: knowns, unknowns, and unknown unknowns. Development and Psychopathology. 2012;24:1, 113-23. 3 World Health Organization. 4 Aboud F & Yousafzai AK. In Black RE, Laxminarayan R, Temmerman M et al., editors.Washington (DC): International Bank for Reconstruction and Development/The World Bank; 2016 Apr 5. 5 Food and Agriculture Organization of the United Nations.

71 | P a g e

micronutrients that may negatively affect growth, development and other relevant

outcomes.1

• (Child) Maltreatment: the abuse and neglect of children under 18 years of age. It includes

all types of physical and/or emotional ill-treatment, sexual abuse, neglect, negligence and

commercial or other exploitation, which results in actual or potential harm to the child’s

health, survival, development or dignity in the context of a relationship of responsibility, trust

or power.2

• Maternal-infant bonding: While widely defined in the literature, a general definition

describes maternal–infant bonding as a process that includes the emotional tie of a mother to

her infant, occurring in the first week or year of an infant's life and that is influenced by

signals and cues from the child as well as the maternal-driven processes.3

• Micronutrients: minerals and vitamins that enable the body to produce enzymes, hormones

and other substances essential for proper growth and development. Consumed in minuscule

amounts, but the consequences of their absence are severe. Iodine, vitamin A and iron are

most important in global public health terms; their lack represents a major threat to the health

and development of populations the world over, particularly children and pregnant women in

low-income countries.4

• Nurturing care: characterized by a caregiving environment that is sensitive to children’s

health and nutritional needs, responsive, emotionally supportive, and developmentally

stimulating and appropriate, with opportunities for play and exploration and protection from

adversities.5

• Nutritional supplementation:

o Multiple vitamin and mineral supplements: multiple micronutrients constitute the

common nutritional supplement provided to young children. Children are often

deficient in many minerals, such as iron and zinc, as well as vitamins. All are critical

for health and growth, and their effects on mental development are becoming clear.9

• Play: a central component of early childhood stimulation and quality parent-child

interactions that is essential to the social, emotional, cognitive, and physical wellbeing of

children beginning in early childhood.6

o Child-led (-driven, -centred) play: play that is directed by the child (though

caregivers may observe or join in), in which children are able to practice decision-

making skills, move at their own pace, and discover their own areas of interest.7

1 Becker et al. Consensus statement of the Academy of Nutrition and Dietetics/American Society for Parenteral and Enteral Nutrition: indicators

recommended for the identification and documentation of pediatric malnutrition (undernutrition). 2015. 2 World Health Organization. 3 Bicking Kinsey C, Hupcey JE . State of the science of maternal-infant bonding: a principle-based concept analysis. Midwifery. 2013;29(12),

10.1016/j.midw.2012.12.019. http://doi.org/10.1016/j.midw.2012.12.019. 4 World Health Organization. 5 Black MM et al. Early childhood development coming of age: science through the life course. Lancet. 2017;389(10064):77-90. 6 Milteer RM. The importance of play in promoting healthy child development and maintaining strong parent-child bond: focus on children in poverty. Pediatrics. 2012;129(1). 7 Ginsburg et al. The importance of play in promoting healthy child development and maintaining strong parent-child bonds. Pediatrics. 2007;19(1).

72 | P a g e

• Positive parenting: consists of five core principles for parents to promote social competence

and emotional self-regulation in children: (1) ensuring a safe, engaging environment, (2)

promoting a positive learning environment, (3) using assertive discipline, (4) maintaining

reasonable expectations, and (5) taking care of oneself as a parent. The emphasis is on

parents learning how to apply these skills to different behavioural, emotional and

developmental issues in children, ranging from common child-rearing challenges (e.g.

toileting, mealtime behaviour, bedtime, behaviour in public) to more intense challenges (e.g.

child aggressive behaviour, fears and anxiety, ADHD difficulties).1

• Psychosocial Stimulation: refers to an external object or event that elicits a physiological

and psychological response in the child.

• Responsiveness: the capacity of the caregiver to respond contingently and appropriately to

the infant’s signals.

• Scaffolding: a concept derived from Vygotsky’s theory of mediated learning, scaffolding is

the process by which someone organizes an event that is unfamiliar or beyond a learner’s

ability in order to assist the learner in carrying out that event.

• Sensitivity: the capacity of the caregiver to be aware of the infant and aware of the infant’s

acts and vocalizations as signals communicating needs and wants.

• Sensitive Discipline: parents’ ability to take into account the child’s perspective and signals

when discipline is required.2

• Stunting: a commonly used indicator of chronic undernutrition, defined as more than two

standard deviations below the age- and gender-specific norm.

• Temperament: an individual’s characteristic mode of responding emotionally and

behaviourally to environmental events. Temperament includes the dimensions of irritability,

activity level, fearfulness and sociability.

• Violent Discipline: actions taken by a parent or caregiver that are intended to cause a child

physical pain or emotional distress as a way to correct behaviour and act as a deterrent.

Violent discipline can take two forms: psychological aggression and physical, or corporal,

punishment. The former includes shouting, yelling and screaming at the child, and addressing

her or him with offensive names. Physical or corporal punishment comprises actions

intended to cause the child physical pain or discomfort but not injuries. Minor physical

punishment includes shaking the child and slapping or hitting him or her on the hand, arm,

leg or bottom. Severe physical punishment includes hitting the child on the face, head or ears,

or hitting the child hard or repeatedly.3

1 Prinz RJ, Sanders MR, Shapiro CJ, Whitaker DJ, Lutzker JR Population-based prevention of child maltreatment: The U.S. Triple P System Population Trial. Prevention Science. 2009;10:1–12. 2 Van Zeijl J et al. Attachment-based intervention for enhancing sensitive discipline in mothers of 1- to 3-year-old children at risk for

externalizing behavior problems: a randomized controlled trial. Journal of Consulting and Clinical Psychology. 2006;74(6): 994-1005. 3 UNICEF.

73 | P a g e

Behaviour Change Techniques

• Information: provision of new information about the link between behaviour and child

development, causes and consequences, and instruction on how to perform the behaviour.1

• Materials: materials that beneficiary families would not normally possess or buy on their

own are provided in order to facilitate behaviour change.

• Media: use of any form of media to bring about behaviour change, including TV

advertisements, flashcards, and organization of role plays and dramas.

• Performance: includes modelling or providing demonstrations, actual rehearsal or practice

of a targeted behaviour in the intervention setting, providing feedback on performance,

contingent rewards, and/or identification of cues to action.

• Problem Solving: includes identifying facilitators and barriers of a targeted behaviour, as

well as solutions to overcoming barriers.

• Social Support: leveraging support from various members of the society/community to

bring about behaviour change; includes motivating peers, family members or authority

figures to encourage parents to engage in behaviour change.

1 Briscoe & Aboud. Behaviour change communication targeting four health behaviours in developing countries: a review of change techniques. Social Science and Medicine. 2012;75(4); 612-21; Aboud FE, Yousafzai AK. Global health and development in early childhood. Annual Review

of Psychology. 2015; 66(1): 433-57.

74 | P a g e

Appendix B: List of Guideline Development Group members

# Name Institution

1 Frances Aboud McGill University, Montreal, Canada

2 Ilgi Ertem Ankara University School of Medicine, Turkey

3 Jane Fisher (Chair) Monash University, Melbourne, Australia

4 Subodh Gupta Dr Sushila Nayar School of Public Health, Maharashtra, India

5 Ghassan Issa Arab Network for Early Childhood Care and Development (ANECD), Beirut,

Lebanon

6 Stewart Kabaka Ministry of Health, Nairobi, Kenya

7 Betty Kirkwood London School of Hygiene & Tropical Medicine, United Kingdom

8 Vibha

Krishnamurthy

Ummeed Child Development Center, Mumbai, India

9 Kofi Marfo Institute for Human Development (IHD), Aga Khan University, Nairobi, Kenya

10 Joerg Meerpohl Cochrane Collaboration, Germany

11 Linda Richter Human Sciences Research Council, University of the Witwatersrand, Johannesburg,

South Africa

12 Fahmida Tofail ICDDR, Bangladesh

13 Mark Tomlinson Department of Psychology, Stellenbosch University, South Africa

14 Susan Walker Tropical Medicine Research Institute, Jamaica

United Nations Partners

15 Pia Britto UNICEF, New York, US

16 Amanda Devercelli World Bank, US

WHO

17 Betzabe Butron Regional Office

18 Teshome Desta IST

19 Martin Weber Regional Office

75 | P a g e

Appendix C: Summary of interventions for all included studies in the systematic

reviews

See PDF supplement - Appendix C Jeong, Franchett, Yousafzai.

Appendix D: GRADE tables and analysis for responsive caregiving interventions (n=17)

Quality of assessments Summary of findings

Child

outcome

No. of

studies

Design Limitations

in study

design and

execution

Inconsistency Indirectness Imprecision Publication

bias

Overall

quality of

evidence

Pooled

effect size

(95% CI)

Cognitive

development

3 RCTs No serious

limitations

Serious limitations

Positive effects

reported by Barrera

et al., 1986; null

effects observed in

remaining studies.

No serious

limitations

Serious

limitations

Pooled effect

size has wide

CI

No serious

limitations

Low 0.26 (-0.14,

0.66); n=1

Language

development

5 RCTs No serious

limitations

No serious

limitations

No serious

limitations

Serious

limitations

Pooled results

have wide CI.

No serious

limitations

Moderate 0.08 (-0.07,

0.23); n=5

Motor

development

2 RCTs No serious

limitations

Serious limitations

Positive impacts

found by Frongillo

et al., 2017, but no

impact found by

Barrera et al., 1992.

No serious

limitations

No serious

limitations

No serious

limitations

Moderate 0.19 (0.12,

0.26); n=1

Socioemotiona

l development

4 RCTs No serious

limitations

No serious

limitations

Serious

limitations

Studies are all

from HICs

Serious

limitations

Pooled results

have a wide

CI.

No serious

limitations

Low 0.14 (-0.03,

0.30); n=4

77 | P a g e

Behaviour

problems

7 RCTs No serious

limitations

No serious

limitations

Serious

limitations

Studies are all

from HICs.

Serious

limitations

Pooled results

have a wide

CI.

No serious

limitations

Low -0.14 (-0.29,

0.002); n=7

Attachment

outcomes

7 RCTs No serious

limitations

Serious limitations

Six studies found

null effects; one

found positive

effects (Cooper et

al., 2009).

No serious

limitations

Serious

limitations

Pooled results

have a wide

CI.

No serious

limitations

Low 0.13 (-0.11,

0.37); n=3

HAZ

1 RCTs No serious

limitations

No serious

limitations

Serious

limitations

This intervention

was a unique

aspect of

responsive

caregiving as it

focused on

responsive

feeding.

No serious

limitations

No serious

limitations


0.16); n=1

WAZ 1 RCTs No serious

limitations

No serious

limitations

Serious

limitations

This intervention

was a unique

aspect of

responsive

caregiving as it

focused on

responsive

feeding.

No serious

limitations

No serious

limitations

Moderate 0.03 (-0.04,

0.10); n=1

78 | P a g e

Caregiving

outcome

No. of

studies

Design Limitations

in study

design and

execution


bias

Overall

quality of

evidence

Pooled

effect size

(95% CI)

Caregiving

knowledge

1 RCT No serious

limitations

No serious

limitations

Serious

limitations

Only HICs

represented.

Serious

limitations

Wide CI.

No serious

limitations

Low 0.29 (-0.01,

0.58); n=1

Caregiving

practices

3 RCTs No serious

limitations

Serious limitations

Two studies found

positive impacts

(Murray et al., 2016;

Barrera et al., 1986);

one study found no

impacts

(Mendelsohn et al.,

2007).

No serious

limitations

Serious

limitations

Wide CI

around the

pooled

estimate.

No serious

limitations

Low 0.53 (-0.10,

1.17); n=2

Caregiver-

child

interaction

8 RCTs No serious

limitations

Serious limitations

Two studies found

no impacts (Barrera

et al., 1986; Van

Zeijl et al., 2006);

remaining studies

found significant

positive impacts.

No serious

limitations

Serious

limitations

Wide CI

around the

pooled

estimate.

No serious

limitations

Low 0.34 (0.15,

0.54); n=6

Caregiver

depressive

symptoms

3 RCTs No serious

limitations

No serious

limitations

No serious

limitations

Serious

limitations

Wide CI

around the

pooled

estimate.

No serious

limitations

Moderate -0.21 (-0.39,

-0.04); n=3

79 | P a g e

80 | P a g e

81 | P a g e

82 | P a g e

83 | P a g e

84 | P a g e

85 | P a g e

Appendix E: GRADE tables and analysis for caregiving interventions to support early learning opportunities

(n=22)


Child outcome No. of

studies

Design Limitations

in study

design and

execution


bias

Overall

quality of

evidence

Pooled effect

size (95%

CI)

Cognitive

development

13 RCTs No serious

limitations

Serious limitations

Positive effects in some

(Muhoozi et al., 2017);

null effects in others

(Norr et al., 2003).

No serious

limitations

Serious

limitations

Pooled effect

size has wide

CI.

No serious

limitations

Low 0.20 (0.01,

0.39); n=8

Language

development

9 RCTs No serious

limitations

Serious limitations

Muhoozi et al., 2017 &

Schwarz et al., 2012:

null effects; Jin et al.,

2007: positive impacts.

No serious

limitations

Serious

limitations

Pooled

results have

wide CI.

No serious

limitations

Low 0.07 (-0.11,

0.24); n=6

Motor

development

7 RCTs No serious

limitations

Serious limitations

Variation in direction

and magnitude of

effects: null effects in

some (Rockers et al.,

2016) and positive

effects in others (Jin et

al., 2007).

No serious

limitations

Serious

limitations

Pooled

results have

wide CI.

No serious

limitations

Low 0.32 (0.12,

0.52); n=5

86 | P a g e

Socioemotional

development

9 RCTs No serious

limitations

Serious limitations

In five out of the six

other studies that could

not be meta-analysed,

there are no statistical

differences.

Serious

limitations

All HICs.

Serious

limitations

Pooled

results have a

wide CI.

No serious

limitations

Very low 0.28 (0.09,

0.48), n=3

Behaviour

problems

8 RCTs No serious

limitations

Serious limitations

Mixed evidence with

some studies finding

differences (e.g. Leung

et al., 2017b; and

Caughy et al., 2004)

versus others finding no

significant differences

(e.g. Goodson et al.,

2000; and Jacobs et al.,

2016).

Serious

limitations

All HICs.

Serious

limitations

Pooled

results have a

wide CI.

No serious

limitations

Very low -0.25 (-0.54,

0.04), n=3

Attachment

outcomes

2 RCTs No serious

limitations

No serious limitations

Both studies reported

significant

improvements.

Serious

limitations

All HICs.

Serious

limitations

Wide CI.

No serious

limitations

Low 0.30 (0.09,

0.51); n=1

HAZ

2 RCTs No serious

limitations


No serious

limitations

Serious

limitations

Pooled

results have a

wide CI.

No serious

limitations

Moderate -0.02 (-0.29,

0.24); n=2


limitations


No serious

limitations

Serious

limitations

No serious

limitations

Moderate 0.05 (-0.10,

0.19); n=2

87 | P a g e

Pooled

results have a

wide CI.

Caregiving

outcome

No. of

studies

Design Limitations

in study

design and

execution


bias

Overall

quality of

evidence

Pooled effect

size (95%

CI)

Caregiving

knowledge

3 RCTs No serious

limitations

Serious limitations

Two studies found

significant

improvements (Jin et

al., 2007; Walkup et al.,

2009); one study found

no effects (Wagner et

al., 2002).

No serious

limitations

Serious

limitations

Zero studies

contributing

to pooled

estimate.

No serious

limitations

Low N.A.

Caregiving

practices

8 RCTs No serious

limitations

Serious limitations

One study (Love et al.,

2005) found statistically

significant

improvements; the other

studies reported no

impact.

Serious

limitations

Only HICs

represented.

Serious

limitations

Over half the

studies have

a small

sample size.

No serious

limitations

Low 0.05 (-0.04,

0.13); n=2

Caregiver-child

interactions

5 RCTs No serious

limitations

Serious limitations

Some studies reported

positive effects (Caughy

et al., 2004; Love et al.,

2005); other studies

reported no impact

(Goodson et al., 2000;

Wagner et al., 2002).

Serious

limitations

Only HICs

represented.

Serious

limitations

Zero studies

contributing

to pooled

estimate.

No serious

limitations

Low N.A.

88 | P a g e

Caregiver

depressive

symptoms

4 RCTs No serious

limitations

No serious limitations No serious

limitations

Serious

limitations

CI around

pooled

estimate is

wide.

No serious

limitations

Moderate 0.07 (-0.08,

0.22); n=2)

89 | P a g e

90 | P a g e

91 | P a g e

92 | P a g e

93 | P a g e

94 | P a g e

Appendix F: GRADE tables and analysis for combined responsive caregiving and the promotion of early learning

interventions (n=42)



studies

Design Limitations

in study

design and

execution


bias

Overall

quality of

evidence

Pooled effect

size (95%

CI)

Cognitive

development

36 RCTs No serious

limitations

Serious limitations


and magnitude of

effects: some studies

have positive impacts

(Aboud et al., 2013);

others had null effects

(Leung et al., 2017a;

Drotar et al., 2008).

No serious

limitations

Serious

limitations

Pooled effect

size has wide

CI.

No serious

limitations

Low 0.45 (0.25,

0.65); n=20

Language

development

17 RCTs No serious

limitations

Serious limitations


and magnitude of

effects: some studies

have positive impacts

(Powell et al., 2004;

Vally et al., 2015);

others had null effects

(Goldfeld et al., 2011).

No serious

limitations

Serious

limitations

Pooled

results have

wide CI.

No serious

limitations

Low 0.38 (0.16,

0.60); n=14

Motor

development

18 RCTs No serious

limitations

Serious limitations

Some studies found

positive impacts

(Yousafzai et al., 2014);

others reported no

impacts (Heinicke et al.,

1999).

No serious

limitations

Serious

limitations

Pooled

results have

wide CI.

No serious

limitations

Low 0.25 (0.09,

0.40); n=13

95 | P a g e

Socioemotional

development

4 RCTs No serious

limitations


No serious

limitations

Serious

limitations

Pooled

results have a

wide CI.

No serious

limitations

Moderate 0.06 (-0.18,

0.28); n=2

Behaviour

problems

7 RCTs No serious

limitations

No serious limitations Serious

limitations

All HICs.

Serious

limitations

Pooled

results have a

wide CI.

No serious

limitations

Low -0.18 (-0.40,

0.04); n=2

Attachment

outcomes

2 RCTs No serious

limitations


All studies found

positive impacts on

attachment outcomes.

No serious

limitations

Serious

limitations

Zero studies

contributing

to pooled

estimate.

No serious

limitations

Moderate N.A.

HAZ

8 RCTs No serious

limitations


No serious

limitations

Serious

limitations

Pooled

results have a

wide CI.

No serious

limitations

Moderate -0.04 (-0.15,

0.07); n=8


limitations


No serious

limitations

No serious

limitations

No serious

limitations

High 0.02 (-0.07,

0.11); n=6

96 | P a g e

Caregiving

outcome

No. of

studies

Design Limitations

in study

design and

execution


bias

Overall

quality of

evidence

Pooled effect

size (95%

CI)

Caregiving

knowledge

7 RCTs No serious

limitations

Serious limitations

Differences in

magnitude of effects;

CIs for Powell et al.,

2004, and Chang et al.,

2015, do not overlap.

No serious

limitations

Serious

limitations

Wide CI

around the

pooled

estimate.

No serious

limitations

Low 0.73 (0.57,

0.89); n=6

Caregiving

practices

18 RCTs No serious

limitations

Serious limitations

Differences in

magnitude and direction

of effects. Some studies

find positive impacts

(Singla et al., 2015;

Yousafzai et al., 2015);

others report null effects

(Chang et al., 2015;

Goldfeld et al., 2011).

No serious

limitations

Serious

limitations

Wide CI

around the

pooled

estimate.

No serious

limitations

Low 0.48 (0.20,

0.76); n=10

Caregiver-child

interactions

12 RCTs No serious

limitations


limitations

Serious

limitations

Wide CI

around the

pooled

estimate.

No serious

limitations


1.10); n=5

Caregiver

depressive

symptoms

9 RCTs No serious

limitations

Serious limitations

Some studies found

significant reductions

(Singla et al., 2015),

No serious

limitations

Serious

limitations

Wide CI

around the

No serious

limitations

Low -0.08 (-0.31,

0.15); n=7

97 | P a g e

while others reported

null effects (Heinecke et

al., 1999).

pooled

estimate.

98 | P a g e

99 | P a g e

100 | P a g e

101 | P a g e

102 | P a g e

103 | P a g e

Appendix G: GRADE tables and analysis for caregiving interventions for responsive caregiving, early learning, or

a combined responsive caregiving and promotion of early learning intervention (n=81)



studies

Design Limitations

in study

design and

execution


bias

Overall

quality of

evidence

Pooled effect

size (95%

CI)

Cognitive

development

52 RCTs No serious

limitations

Serious limitations

Variation in magnitude

and direction of effects:

positive effects

observed in some

studies and null effects

in others.

No serious

limitations

Serious

limitations

Pooled effect

size has wide

CI.

No serious

limitations

Low 0.37 (0.22,

0.52); n=29

Language

development

31 RCTs No serious

limitations

Serious limitations

Positive impacts for

some (Vally et al.,

2015; Powell et al.,

2004) and null effects

for others (Chang et al.,

2015; Guttentag et al.,

2014).

No serious

limitations

Serious

limitations

Pooled

results have

wide CI.

No serious

limitations

Low 0.24 (0.11,

0.36); n=25

Motor

development

27 RCTs No serious

limitations

Serious limitations

Positive impacts in

some studies (Yousafzai

et al., 2014); null effects

in others (Gardner et al.,

2005).

No serious

limitations

No serious

limitations

No serious

limitations


0.37); n=19

104 | P a g e

Socioemotional

development

17 RCTs No serious

limitations

Serious limitations

No statistical

differences in many

studies that could not be

meta-analysed.

No serious

limitations

Serious

limitations

Pooled

results have a

wide CI.

No serious

limitations

Low 0.15 (0.04,

0.27); n=9

Behaviour

problems

22 RCTs No serious

limitations

Serious limitations

Mixed findings: many

studies that could not be

meta-analysed found

null impacts.

No serious

limitations

Serious

limitations

Pooled

results have a

wide CI.

No serious

limitations

Low -0.17 (-0.28,

-0.06); n=12

Attachment

outcomes

11 RCTs No serious

limitations

Serious limitations

Positive effects in some

studies (Cooper et al.,

2009; Guedeney et al.,

2013); null effects in

others (Kalinauskiene et

al., 2009).

No serious

limitations

Serious

limitations

Pooled

results have a

wide CI.

No serious

limitations

Low 0.23 (0.07,

0.38); n=4

HAZ

11 RCTs No serious

limitations


limitations

No serious

limitations

No serious

limitations

High -0.02 (-0.10,

0.07); n=11


limitations


No serious

limitations

No serious

limitations

No serious

limitations

High 0.03 (-0.02,

0.08); n=9

Caregiving

outcome

No. of

studies

Design Limitations

in study

design and

execution


bias

Overall

quality of

evidence

Pooled effect

size (95%

CI)

Caregiving

knowledge

11 RCTs No serious

limitations

Serious limitations

No serious

limitations

Serious

limitations

No serious

limitations

Low 0.68 (0.51,

0.85); n=7

105 | P a g e

Differences in

magnitude of effects;

CIs for Powell et al.,

2004, and Chang et al.,


Wagner et al., 2002,

reported null effects.

Wide CI

around the

pooled

estimate.

Caregiving

practices

29 RCTs No serious

limitations

Serious limitations


null effects (Wasik et

al., 1990; Goldfeld et

al., 2011); others found

significant

improvements (Singla et

al., 2015; Yousafzai et

al., 2015).

No serious

limitations

Serious

limitations

Wide CI

around the

pooled

estimate.

No serious

limitations

Low 0.44 (0.21,

0.67); n=14

Caregiver-child

interactions

25 RCTs No serious

limitations

Serious limitations


null effects (Van Zeijl et

al., 2006; Wagner et al.,

2002); others found

significant

improvements (Murray

et al., 2016; Guttentag

et al., 2014).

No serious

limitations

Serious

limitations

Wide CI

around the

pooled

estimate.

No serious

limitations

Low 0.54 (0.30,

0.78); n=11

Caregiver

depressive

symptoms

16 RCTs No serious

limitations

Serious limitations

Some studies found

significant reductions

(Baker-Henningham et

al., 2004); others found

no effects (Heinicke et

al., 1999).

No serious

limitations

Serious

limitations

Wide CI

around the

pooled

estimate.

No serious

limitations

-0.07 (-0.22,

0.07); n=12

106 | P a g e

107 | P a g e

108 | P a g e

109 | P a g e

110 | P a g e

111 | P a g e

112 | P a g e

Appendix H: GRADE tables and analysis for caregiving interventions to support healthy child socioemotional and

behavioural development (n=10)



studies

Design Limitations

in study

design and

execution


bias

Overall

quality of

evidence

Pooled effect

size (95% CI)

Cognition

development

1 RCT No serious

limitations

No serious

limitations

Serious

limitations

n=1 study

from HIC.

Very serious

limitations

n=1 study;

pooled ES

NA.

No serious

limitations

Very low No pooled ES.

From Caldera et

al., 2007:

Adjusted ES for

BSID Mental

Development

Index = 0.29;

p<0.05

Motor

development

1 RCT No serious

limitations

No serious

limitations

Serious

limitations

n=1 study

from HIC.

Very serious

limitations

n=1 study;

pooled ES

NA.

No serious

limitations


From Caldera et

al., 2007:

Adjusted ES for

BSID

Psychomotor

Development

Index = 0.19;

p=0.16

Prosocial

behavior/

Socioemotional

development

1 RCT No serious

limitations

No serious

limitations

Serious

limitations

n=1 study

from HIC.

Very serious

limitations

n=1 study;

pooled ES

NA.

No serious

limitations


From Barlow et

al., 2015:

Adjusted ES for

ITSEA= 0.14;

p=0.09.

Behaviour

problems

10 RCT No serious

limitations

No Serious

limitations

Serious

limitations

No serious

limitations

No Serious

limitations

Moderate -0.02

(-0.07, 0.02)

113 | P a g e



studies

Design Limitations

in study

design and

execution


bias

Overall

quality of

evidence

Pooled effect

size (95% CI)

I2 = 9.9%

95% CIs are

overlapping.

Additional n=5

studies:

n=1: no effects; n=1:

significant

reductions in mean

scores for child

behaviour problems;

n=3: significant

reductions on some

domains but not all.

n=10 studies

from HICs.

n=5

114 | P a g e


Caregiver

outcome

No. of

studies

Design Limitations in

study design

and dxecution


bias

Overall

quality of

evidence

Pooled

effect

size

(95%

CI)

Caregiving

practices

8 RCT No serious

limitations

Serious limitations

n=4 no effects; n=3

mixed effects by

subscale; n=1 significant

improvements.

Serious

limitations

n=8 studies

from HICs.

Serious

limitations

No serious

limitations

Low 0.01

(-0.04,

0.06)

n=2

Caregiver

mental

health

4 RCT No serious

limitations

Serious limitations

95% CIs for Shaw et al.,

2009 & Hiscock et al.,

2008 are not

overlapping.

Shaw et al., 2009 and

Barlow et al., 2015

found significant

declines; Hiscock et al.,

2008 & Hiscock et al.,

2018 found no effects.

Serious

limitations

n=4 studies

from HICs.

No Serious

limitations

All studies

>85/arm

95% CI for

pooled ES is

relatively tight.

No serious

limitations

Low -0.05

(-0.11,

0.01)

n=3

Caregiving

knowledge

2 RCT No serious

limitations

Serious limitations

Barlow et al., 2015

found significant

improvements; Caldera

et al., 2007 found no

effects.

Serious

limitations

n=2 studies

from HICs.

Serious

limitations

Sample size is

adequate (>85 per

group)

Data cannot be

meta-analysed.

No serious

limitations

Very Low

Caregiver-

child

interactions

5 RCT No serious

limitations

Serious limitations

n=2 found no effects;

n=1 found significant

Serious

limitations

Serious

limitations

No serious

limitations

Very Low 0.14

(-0.07,

0.34);

n=1

115 | P a g e

improvements; n=1

found mixed results

across measures.

n=5 studies

from HICs.

Sample size is

adequate (>85 per

group)

Only n=1 study

included in pooled

results.

Self-efficacy 3 RCT No serious

limitations

Serious limitations

Breitenstein et al., 2012

& Caldera et al., 2007

found significant

improvements; Gross et

al., 2009 found no

effects.

Serious

limitations

n=3 studies

from HICs.

Serious

limitations

Sample size is

adequate (>85 per

group)

Data cannot be

meta-analysed

(unadjusted

means and SDs

are not presented).

No serious

limitations

Low

116 | P a g e

117 | P a g e

Appendix I: GRADE tables and analysis for integrated caregiving and nutrition interventions (n=18)

Combined nutrition and caregiving interventions versus standard of care



studies

Design Limitations

in study

design and

execution


bias

Overall

quality of

evidence

Pooled effect

size (95%

CI)

Cognitive

development

14 RCTs No serious

limitations

Serious limitations


and direction of effects;

some studies have null

effects (Rockers et al.,

2016; Nahar et al.,

2012), while others

have positive effects

(Aboud et al., 2013;

Grantham-McGregor et

al., 1991).

No serious

limitations

Serious

limitations

Pooled effect

size has wide

CI.

No serious

limitations

Low 0.57 (0.32,

0.82); n=13

Language

development

10 RCTs No serious

limitations

Serious limitations


and direction of effects:

Muhoozi et al., 2017,

found negative impacts;

Aboud et al., 2013 &

Yousafzai et al., 2014

found positive impacts.

No serious

limitations

Serious

limitations

Pooled

results have

wide CI.

No serious

limitations

Low 0.40 (0.17,

0.63); n=10

Motor

development

10 RCTs No serious

limitations

Serious limitations

No serious

limitations

Serious

limitations

No serious

limitations

Low 0.4 (0.26,

0.53); n=10

118 | P a g e


and direction of effects.

Nahar et al., 2012, and

Vazir et al., 2013 found

null effects, whereas

others found positive

impacts (Yousafzai et

al., 2014; Muhoozi et

al., 2017).

5 of the 10

studies have

small sample

sizes.

Pooled effect

size has wide

CI.

Socioemotional

development

2 RCTs No serious

limitations

Serious limitations

Yousafzai et al., 2015 –

no impacts; Muhoozi et

al., 2017 – positive

impacts.

No serious

limitations

Serious

limitations

Pooled effect

size has wide

CI.

No serious

limitations

Low 0.09 (-0.11,

0.30); n=2

HAZ

9 RCTs No serious

limitations

Serious limitations

Nahar et al., 2012, and

Helmizar et al., 2017,

found negative impacts;

others found null

effects.

No serious

limitations

Serious

limitations

Pooled

results have

wide CI.

No serious

limitations

Low -0.13 (-0.31,

0.05); n=9


limitations


No serious

limitations

No serious

limitations

No serious

limitations

High 0.06 (-0.02,

0.13); n=7

WHZ 6 RCTs No serious

limitations


No serious

limitations

Serious

limitations

Pooled

results have

wide CI.

No serious

limitations


0.34); n=6

119 | P a g e

Combined nutrition and caregiving interventions versus caregiving interventions



studies

Design Limitations

in study

design and

execution


bias

Overall

quality of

evidence

Pooled effect

size (95%

CI)

Cognitive

development

7 RCTs No serious

limitations

Serious limitations

Yousafzai et al., 2014:

negative effects;

Gardner et al., 2005 &

Grantham-McGregor et

al., 1991: positive

effects.

No serious

limitations

Serious

limitations

Pooled effect

size has wide

CI.

No serious

limitations

Low 0.10 (-0.12,

0.32); n=6

Language

development

10 RCTs No serious

limitations


No serious

limitations

Serious

limitations

Pooled

results have

wide CI.

No serious

limitations

Moderate 0.01 (-0.09,

0.10); n=6

Motor

development

10 RCTs No serious

limitations

Serious limitations

Gardner et al., 2005,

found positive impacts;

other studies (Yousafzai

et al., 2014; Nahar et

al., 2012a) found no

impacts.

No serious

limitations

Serious

limitations

Pooled

results have

wide CI.

No serious

limitations

Low 0.18 (-0.06,

0.42); n=6)

120 | P a g e

Socioemotional

development

1 RCTs No serious

limitations

Serious limitations

No serious

limitations

Serious

limitations

Pooled

results have

wide CI.

No serious

limitations

Low 0.11 (-0.04,

0.26); n=1

HAZ

9 RCTs No serious

limitations

Serious limitations

Nahar et al., 2012a,

found negative effects;

Aboud & Akhter, 2011,

and Yousafzai et al.,

2014, found null effects.

No serious

limitations

Serious

limitations

Pooled

results have

wide CI.

No serious

limitations

Low -0.21 (-0.60,

0.19); n=4


limitations


No serious

limitations

Serious

limitations

2 of the 3

studies have

small sample

sizes.

No serious

limitations

Moderate 0.07 (-0.04,

0.17); n=3


limitations


No serious

limitations

Serious

limitations

Wide CI; 3 of

the 4 studies

have small

sample sizes.

No serious

limitations


0.29); n=4

121 | P a g e

Combined nutrition and caregiving interventions versus nutrition interventions



studies

Design Limitations

in study

design and

execution


bias

Overall

quality of

evidence

Pooled

effect size

(95% CI)

Cognitive

development

10 RCTs No serious

limitations

Serious limitations


and magnitude of

effects: Lozoff et al.,

2010, found positive

effects; Nahar et al.,

2013, found null

effects.

CIs for Lozoff et al.,

2010, and Nahar et al,


No serious

limitations

Serious

limitations

5 of the 9

studies have

small sample

sizes.

Pooled effect

size has wide

CI.

No serious

limitations

Low 0.45 (0.22,

0.67); n=9

Language

development

6 RCTs No serious

limitations

No serious

limitations

No serious

limitations

Serious

limitations

3 of the 6

studies have

small sample

sizes.

No serious

limitations


0.28); n=6

Motor

development

9 RCTs No serious

limitations

No serious

limitations

No serious

limitations

No serious

limitations

No serious

limitations

High 0.14 (0.07,

0.22); n=9

122 | P a g e

Socioemotional

development

1 RCTs No serious

limitations

No serious

limitations

No serious

limitations

Serious

limitations

Pooled results

have wide CIs.

No serious

limitations

Low -0.09 (-0.24,

0.07); n=1

HAZ

4 RCTs No serious

limitations

Serious limitations

Variation in

magnitude and

direction of effects:

Nahar et al., 2012,

found negative

impacts; Menon et

al., 2016, found

positive impacts.

No serious

limitations

Serious

limitations

2 of the 4

studies have

small sample

sizes.

Pooled results

have wide CI.

No serious

limitations

Low -0.42 (-0.85,

0.01); n=4


limitations

No serious

limitations

No serious

limitations

Serious

limitations

2 of the 4

studies have

small sample

sizes.

No serious

limitations

Moderate 0.06 (-0.02,

0.14); n=4


limitations

Serious limitation

Helmizar et al., 2017,

found positive

impacts, while the

other studies did not.

No serious

limitations

Serious

limitations

3 of the 5

studies have

small sample

sizes.

No serious

limitations

Low 0.17 (-0.04,

0.38); n=5

123 | P a g e

Combined nutrition and caregiving interventions versus standard of care

Cognitive Development

Language

124 | P a g e

Motor

Socioemotional

125 | P a g e

HAZ

WAZ

126 | P a g e

WHZ

127 | P a g e

Combined nutrition and caregiving interventions versus nutrition


Language

128 | P a g e

Motor

Socioemotional

129 | P a g e

HAZ

WAZ

130 | P a g e

WHZ

131 | P a g e

Combined caregiving and nutrition interventions versus caregiving


Language

132 | P a g e

Motor

Socioemotional

133 | P a g e

HAZ

WAZ

134 | P a g e

WHZ

Documents

World Health Organization Recommendations on Caregiving